Recent trends in calciner design

1 Introduction
The cement industry has become significantly more energy efficient during the last 30 years with new plants having the standard features of cyclone preheater systems with precalciner technology, short kilns burning 35 to 45  % of the total fuels as well as incorporating multi-stage clinker cooling systems. Further improvements through the use of alternative fuels and raw materials (AFR) have led to cost reduction measures, as modifications to the existing plants are sought and new plants are being built making provision for AFR. The variation in the availability and properties of the AFR has generated a number of combinations for the utilization of these fuel types as and when they became available. However, their full potential has never been reasonably explored in the absence of sufficient under­standing of their combustion, emission and mineral reaction interactions taking place inside the kiln and precalciner and hence there is a danger of incurring increased AFR preparation costs as well as excessive capital costs associated with increase in calciner size. The aim of the present paper, therefore, is to highlight the advantages of gaining more information on mixing, combustion, emission and calcination within a precalciner, prior to opting for higher cost solutions which may or may not deliver the expected performance.
 
Detailed measurement and experimentation at cement plants were, and remain, most of the time, both very difficult and prohibitively expensive as kiln upsets and short “outage” spell(s) do not allow the process engineer to introduce and study many plausible modifications. Therefore, design tools and improvements to the kiln/precalciner, and other combustion-based facilities have evolved slowly. Initially, heat and mass-based models, frequently employing empirically-based correlations, were used during the early part of the 20th Century. The 1950s saw the increased use of cold flow physical models as a design aid. Higher particle loading also made it difficult to reliably collect gas samples and take temperature and velocity measurements in precalciners. However, even if data were collected, the results only encompass the qualitative flow pattern without quantification of the temperature, gas/particle types and emissions.
 
In a modern analysis of precalciner/kiln systems, one would clearly prefer to visualize the effect of a change under combustion conditions prior to its implementation, particularly when complex combinations of alternative fuels and raw materials are used. Therefore, in place of expensive and yet limited experimentation it is desirable to solve, to a useful level of accuracy, detailed combustion and mineral equations to identify solid-phase trajectories and gas-phase pathways; highlighting potential build-up areas, pollutant formation and lower oxygen regimes and avenues of production enhancement. This can only be achieved if detailed computational fluid dynamics (CFD) calculations are carried out in conjunction with mineral reactions. A recently developed mineral-interactive CFD technique, MI-CFD, has been thoroughly validated against laboratory-scale to industrial-scale data and applied to more than 60 precalciners. The details of the MI-CFD approach are presented elsewhere [1–8], however it is also important to understand the information required from the plant, in order to build a precalciner MI-CFD model, as briefly summarized below:
 
2    Plant data input for MI-CFD calculations:
By way of example, a typical base case exercise targeting a precalciner would initiate with a request to the plant for the following information:
 
(1)    Detailed drawings of the riser duct and precalciner showing all the inlets and outlets including the locations of coal/petcoke burners, meal inlet chutes, splash plates, waste fuel chute, etc;
(2)    Coal/petcoke/waste fuel chemical compositions, particle/chip size distributions, burner feed rates, fuel transport air mass flow rates and corresponding temperatures;
(3)    Hot meal particle size distribution, feed rate at each injection location and hot meal analysis at the calciner entry locations and at the calciner exit;
(4)    Detailed dimensions and orientations of the fuel burners;
(5)    Tertiary air flow rate and temperature at each injection ­location;
(6)    Kiln inlet/backend combustion gases compositions (i. e., O2, CO, NOx, CO2), temperature and estimated dust loading and kiln main burner fuel(s) feed rate(s) and air flows and corresponding temperatures;
(7)    Finally, precalciner exit data: temperature, oxygen, extent of char particle carbon burnout, hot-meal calcination levels, pressure drop between kiln inlet and precalciner outlet.
 
Most of the process data is readily available in modern PIMS and laboratories, however, extreme care is needed to ensure that accurate up-to-date drawings are provided.
3    Application of MI-CFD
As an example of the use of the MI-CFD technique Figure 1 reveals the effect of burner location and momentum on the final burnout of the coal particles as a function of their trajectories through the oxygen rich and oxygen lean regions of a precalciner. The precalciner was originally converted from gaseous fuel to coal firing. Particle burnout is clearly higher for those particles, which travel within higher oxygen concentration regions.
 
From the figure it is clearly visible that the tertiary air with a higher oxygen content does not fully mix with the hotter kiln gases with a lower oxygen content until reaching the calciner’s exit, having less than 2 s residence time. The calciner exit on the opposite side to the tertiary air also contributes to the flow stratification by “short-circuiting” some of the tertiary air, which leaves the calciner without mixing with the oxygen-starved fuel particles (Fig. 2). Coal and meal particles only have up to a 1.5 second residence time; with particle burnout and meal calcination levels 82  % and 88  %, respectively.
 
An improvement to the design was proposed by the suppliers, to increase the riser duct height above the tertiary air duct and below the upper expansion area of the calciner (Fig. 3). It was predicted that the change would slightly improve the oxygen diffusion into the riser duct gases but still mixing is delayed until very near the calciner’s exit and hence is not the solution needed for better burnout. In order to fully utilize the calciner potential, either changes to the tertiary air configuration and/or calciner outlet location or burner relocation/enhancement need to be investigated. However, it is difficult to find the best combination either through plant trails as some of the modifications would be quite expensive and may not be realized within a single outage or through simple calculation procedures, as complex non-linear variables cannot be decoupled and any such attempt would be unreliable. Hence, a detailed MI-CFD approach could be used to identify the optimum required modification(s).
 
A detailed study comprising optimization of input and boundary conditions, taking into consideration the number of burners, petcoke/coal burners fuel split, burner momentum and meal split was performed which resulted in improvements of up to 93 and 94  % burnout of coal and calcination of meal particles, (as compared with 82  % and 88  %), respectively. However, it was found that a more rigorous mixing regime at the entry location of the tertiary air is needed in order to reduce flow stratification within the short residence time in the calciner. A conceptual design of the tertiary air oppose-jet configuration with radial oxygen distribution is shown in Figure 4. In the modified design, tertiary air is split into two jets which are allowed to mix with the riser flow through an opening all along three calciner walls. Two coal burners are located at the axis of the tertiary air inlets, through which petcoke/coal particles are instantly ignited and burnt products are radically mixed with the riser flow. A number of simulations were carried out to design the coal burners so that particles do not hit the refractory walls and enter into the riser duct flow without producing higher temperature regions near the refractory walls. Meal balancing was also carried out to promptly quench the higher temperature regions. Significant improvements have been observed in petcoke particle burnout and calcinations levels. When tertiary air duct modifications, similar to the case described, were implemented with a “Capex” of less than 1.5 million €, with further improvement in burner and calcination (98  % and 95  %, respectively).
 
The created hot-spot also reduced the exit NO emissions through the higher reduction of kiln-formed NOx through CHi radicals, thereby leading to higher reburn NOx reduction efficiency as a result of higher localized temperature within the reburn zone. Improved mixing also reduced CO emissions due to higher destruction of the kiln-formed CO and less formation of CO within the calciner through the earlier consumption of evolved volatiles and completion of char burnout.
 
4    Deposit built-up and CO emissions:
Figure 5 shows a sketch of a low NOx calciner in which an in-line precalciner joins the tertiary air duct post combustion. The calciner is fired with pulverized coal and a mixture of tyre chips and diaper cubes in tertiary air and the riser ducts. Tire chips and diaper cubes are injected into the riser duct whereas wood chips and plastic disks are burned in the tertiary air duct. However, it was observed that a proportion of the tire chips, and a smaller fraction of the diaper cubes, dropped down into the riser and tertiary air ducts causing blockages. The blockage problem was more severe in the tertiary air duct due to the elbow at the bottom of the vertical run. Several unsuccessful attempts were made to reduce the build-up, for example, construction of refractory restrictions to accelerate the flow within the tertiary air riser duct. MI-CFD enabled the source of the problem to be traced to the aerodynamics at the join between the riser and tertiary air ducts. An area of recirculation is generated by the horizontal platform connecting the two ducts at the join (Fig. 6). Solid matter settled on this platform. Over time, the build-up would reach a size where chunks would fall from it blocking in particular the tertiary air duct. The straightforward remedy was to construct a refractory prism on the platform, aerodynamically smoothing the join and eliminating the zone of recirculation (Fig 7). Tire and diaper feed rates have been increased without problems from 30  % to almost 50  % of the calciner thermal input. MI-CFD simulations have also shown that even higher substitution of coal fuel is possible through the injection of plastics and wood chips into the tertiary air duct.
 
The higher substitution of waste fuels resulted in CO emissions closer to the permitted limits, which were mainly due to the flow stratification within the upper duct joining the tertiary air and riser ducts (Fig. 8). The upper duct then enters into the pyro­clone where further mixing takes places but the temperatures there are in the range of 950 °C which quickly drop to 850 °C and therefore not all the formed/transported CO is oxidized within the short residence time of less than half a second. In general, oxidation of CO is very slow at these temperatures, as has been calculated with the help of a simple chemical kinetic model (Fig. 9).
 
The introduction of a pair of mixing air jets (Figs. 10–11) does not produce the desired effects as it produces higher velocity flow in the region where tires and diapers are injected which forces their volatiles to shift into the fuel-rich region of the upper section. Nevertheless, it is perceived that setting up mixing air jets within the combined upper section, where a significant separation between the oxygen-rich and fuel-rich streams persists, can result in better combustion of fuel products and reduction of CO emissions. When these jets were moved upwards, CO emissions were reduced about 20  % within a shorter residence time through elimination of fuel-rich pockets. Further reductions were predicted, of the order of 50  %, with increase of the mixing air fan momentum.
 
5    Calciners with cyclonic tertiary air inlet
In another precalciner (Fig. 12), tertiary air is introduced through a cyclonic motion which impart a swirl to its flow and as a consequence, most of it remains closer to the walls, whereas the riser duct flow moves closer to the calciner axis in the upward direction. Under these conditions, most of the petcoke particles are entrained by the higher upward velocity riser duct gas flow and react with oxygen under vitiated air conditions. This leads to higher CO formation through unreacted volatiles present within the inner stream with lower oxygen content. CO emissions values were reduced through lowering the petcoke burners closer to the tertiary air entrance, from 1250 to 600 vppm.
 
6    Building “Big”
It has been observed that in some of the earlier precalciners, although sufficient gas residence time had been made available, height restrictions through the incorporation of bends produce flow stratification. Consequently, little advantage is gained from the extra gas residence time, as shown in the following examples. In all these precalciner configurations, although the gas residence time is from 3 to 8 seconds (from left to right), due to the sudden path change after the bend, flow mixing is suppressed and unmixed streams accelerate through the inner path, whereas flow through the outer section slows down, thereby creating flow separation conditions.
 
For these types of precalciners, it turned out to be difficult to find a general solution and each problem had to be considered separately. For example, a supplier for a plant (Fig. 13) suggested the option to either increase the height of its riser duct (Figs. 14–15) in order to increase gas/particle residence time or to incorporate a tertiary air duct for providing more oxygen and hence increasing the plant clinker production and petcoke usage, while reducing build-up problems.
Simulations showed that lowering the burner position had achieved a little extra residence time for the petcoke within the chamber, however, the high momentum kiln flow in the lower section had minimized the benefit by accelerating the petcoke particles and also pushing them closer to the walls. The low volatile matter content of petcoke is, as expected, the reason for the slow ignition shown here. The combined effect of lowering the petcoke burners and extending the height of the precalciner by 3 m had only marginal effect on char burnout and calcination levels as shifting the burners also comes immediately under the influence of calciner meal particles which “quench” the flame and hence marginal benefit is gained later on with a slight increase in the petcoke burnout, of the order of 4 and 5  % in the petcoke burnout and hot meal calcination level respectively.
 
The proposed upgrade was thoroughly assessed by means of MI-CFD simulations. In the light of previous experience at other plants, it was decided to improve the performance of the petcoke burner by introducing a higher momentum air steam, which resulted in increasing the burning of petcoke by 11 % and also increased the calcination level by 6  %. An increase that was achieved mainly through the penetration of the particles into the riser duct flow and improved mixing in the oxygen-rich region. Therefore, the plant implemented the higher momentum burner option, which avoided the high CAPEX route.

An example of an in-line precalciner with cyclone combustion chamber is shown in Figure 16. The precalciner consists of a swirl chamber fitted with two coal/petcoke burners. Tyre chips were fed from a bucket conveyor plus chute system, open to the atmosphere. The injection location at the top of the calciner was offset from the calciner centreline. The ingress of false air as a result of the arrangement was considerable and this had the effect of quenching the coal flame to the extent that it was difficult to distinguish separate combustion regimes for each of the two coal burners. A series of computer simulations were performed primarily to establish the effect of minimizing the false air and to assess the effect of moving the tire chip injection location to the centreline of the calciner.
 
Reduction of the false air by 50  % resulted in a more uniform calciner temperature and the re-emergence of distinct flames for each burner, evident in the predictions of Figure 17. This ­figure shows the predicted calciner temperature contours showing the effect of reducing the false air ingress associated with the tyre chip feed. The degree of tire chip devolatilization achieved at the level of the kiln riser increased from 30 to 98  %. Since the volatile content of tires is high, i.  e. approximately 60  %, and the volatile gases readily combust, this translates to a very significant improvement in tire chip burnout levels. Halving of the false air was achieved at the plant through the installation of a secondary tire chip drop-gate and sequential opening of this and the primary gate.
 
The MI-CFD exploration of the effect of relocating the tire chip injection position to the centreline of the calciner revealed no improvement in tire chip burnout. The plant therefore ­abandoned this proposal, reaping a considerable cost saving in so doing.

Given the success of the primary MI-CFD studies, supplementary studies followed. One explored the effect of increasing the kiln gas flow at the expense of the tertiary airflow in order to achieve higher kiln throat velocities with the aim of reducing chip fall-out. However, the consequent redistribution of the oxygen was shown to have the undesirable effect of reducing the level of oxygen available for coal/petcoke particle ignition and combustion, which in turn reduced the temperatures and tire devolatilization levels within the swirl chamber. Specifically, the degree of tire devolatilization at the throat dropped to 45  %, the coal/petcoke mixture char burnout dropped from some 90  % to 80  % at the precalciner exit and the degree of calcination at the precalciner exit deteriorated from 95  % to 89  %.
 
Another improvement suggested by the plant was to relocate the coal/petcoke burners within the tertiary air ducts, closer to the hot meal injection location. The simulations showed that the early calcination of hot meal particles prematurely consumed heat, leaving little heat for tire chip devolatilization. The degree of calcination at the precalciner exit was found to be 95  %, but the degree of tire chip devolatilization at the throat reduced by 30  %. Yet another investigation targeted diverting the tertiary air towards the centre of the swirl chamber. This had some beneficial effect, but insufficient to warrant the cost of the modification.
A similar example is another retrofit where a cyclone combustion chamber was incorporated into an in-line precalciner in order to introduce AFR. In this configuration, tertiary air introduced tangentially does not mix well with the coal and AFR particles injected from the chamber axis. Flow stratification persists in the upper calciner section where both in-line riser duct gases mix with the combustion products formed within the CC chamber. Generally, it should be noted that while the concept of CC chambers to improve combustion is good, their individual design can lead to combustion issues, which need to be addressed mainly through burner optimization and curbing the flow stratification.
 
7    Long in-line precalciners:
The next example shows a single tertiary air duct precalciner (Fig. 18) with residence time comparable to some of the long and inclined precalciners presented (Fig. 13). The burner arrangement and tertiary air injection scheme consumes oxygen much faster when compared with the other configurations and long calciner residence time, which help in obtaining higher particle burnout of petcoke particles reaching 99.99  %, near the calciner exit, with calcination levels of 96  % under most operating conditions.
 
An inclined precalciner (Fig. 19) with a similar residence time was modelled and showed the typical signs of poor particle burnout and lower calcination levels. To address the flow stratification problem for those configurations where tertiary air/burner modifications are not feasible, another approach of injecting high momentum air streams could be successfully applied. An MI-CFD study was undertaken in detail and as a remedy a mixing air fan (MAF) was installed in order to accelerate the gas stream mixing. Subsequent to several optimization simulations, it was established that it would be necessary to supply at least 50  % momentum of the rising flow, through the opposing jets in order to impart sufficient cross-flow strength for mixing of the fuel-rich fuel-lean regions.
 
To find an optimized combination of the above parameters, economical and practical considerations, i.  e. minimizing airflow and compressor pressure, and nozzle size specifications, were taken into account in formulating the MI-CFD simulations (Fig. 20). An opposed-jet arrangement provides excellent mixing within the central region of the calciner, but produces very poor mixing within the regions closer to the calciner wall. The same effects are observed when two jets are introduced at two different levels but perpendicular to each other. It has been found that the best arrangement, which provides good mixing within both the central region and the regions closer to the calciner wall, is a pair of opposed jets injected at an angle of 60 º to the normal. However, the introduction of ambient air also leads to increased fuel consumption, therefore as a rule of thumb, velocities of up to 250 m/s and flow rates of 10  % of the precalciner stoichiometric air were considered to be acceptable.
 
Results of the simulations are presented in Figures 21–25, where a “Base case” without MAF and the average operating conditions of the plant is compared with two optimization cases. For Cases 1 and 2, 2 and 4  % of the calciner stoichiometric air was used (equally divided into two jets) with injection velocities of 200 m/s. Improvement in the precalciner performance as compared with the base case is summarized in Table 1.
The two identical opposed jets would impinge on each other in the centre and form a thin disk-like shape which helps in the mixing of the flow, however under higher velocity cross-flow conditions, i.e., in the precalciner, these jets are deflected by the upward flow and in most cases do not hit each other. As a result, they are less effective in promoting the mixing as compared with their impact where cross-flow velocities are low, i.e., near the kiln backend outlet, where velocities are much lower. In order to improve MAF performance and reduce NOx emissions, an inclination angle of 60 degrees to the normal was used along with impregnating air with NH4OH solution (NH3/NO ratio of 1). Results show (Figs. 20–25) that MAF2 had bigger impact on the mixing, as compared with MAF1, due to its higher momentum, as seen from the petcoke and meal particle trajectories (Figs. 24–25) which come under direct influence of the two jets and spread out more in the radial direction. Both the particle burnout and calcination levels increase for the two MAF cases as compared with the base case. NO emission concentrations were also reduced from 910 to 630 vppm, due to improved dispersion of NH3 to react with the NO being produced downstream. MAF’s potential in calciner, however, has not fully been explored. The technique applied showed the benefits of reducing build-up with less frequent cleaning of the feed-shelf area and the use of air cannons associated with a small sacrifice of thermal efficiency.
 
8    Conclusions
Simulation methodology using computational fluid dynamics (CFD) has come far since its first tentative engineering usage in the aerospace industry over 30 years ago, to the extent that it can now be used to simulate complex processes such as those encountered in a cement plant. A particular strength is its ability to directly couple mineral transformation reactions with combustion calculation procedures (MI-CFD) in order to predict more accurate and realistic temperature and gas species composition domains, which are essential, in order to establish realistically the flow and temperature fields and gas species concentrations. With regard to addressing the problems of existing calciners and the expectations as regards future calciners, ­
MI-CFD has the potential to simulate the combustion aerodynamics, the pollutant emission, as well as the calcination levels and to offer the most efficient and cost effective solutions.
 
The examples presented are demonstration examples of various precalciner designs, extending from earlier designs to the most recent ones, constructed by leading plant manufacturers. Based on our MI-CFD simulations and the plant feedback, we have established that moderate to excellent improvements have been achieved, depending on the calciner’s mixing characteristics with minor “Capex” modifications, including relocation of fuel ports, burner enhancement, tertiary air and meal balancing and adjustments. A step which is highly recommended for all existing plants is that an increase in calciner throughput may be achieved through the incorporation of a precombustion chamber, in particular if large-sized and higher density alternative fuels are to be used, as compared with extending the equivalent residence time of the calciner. However, as far as the current trend of building “big” calciners is concerned, which allow residence times of 8–10 seconds, these are not really necessary if, in the design, more attention is given to the early mixing and combustion of the AFR, coal, petcoke, tertiary air and kiln gases. In an upgrade and/or implementation of new precalciner designs, it is more cost effective to first assess the mixing pattern of the precalciner and to establish where the various reaction zones are and how they interact with the help of simulation tools, such as are presented in this paper, so that the best modifications and improvements can be selected and installed.
 
Acknowledgements
The authors gratefully acknowledge the collaboration and ­financial support of our clients as well as their permission to present these MI-CFD specific mathematical modelling results.