BACKGROUND

[0001] Europe has a significant industrial minerals industry, accounting for an estimated turnover of €10 billion annually. As this industry is sensitive to public acceptance, it continuously seeks to improve the sustainability of minerals extraction. Its aim is to ensure that all stages of extraction are performed in an optimal manner from economic, environmental and social perspectives. To achieve these goals, innovative technologies and approaches need to be adopted along the entire mineral value chain, starting with exploration and extraction and extending to re-use and recycling. In particular, the connection between mineralogy and production performance has to be taken into account for economic and sustainable processing of the ore body. Smart mining concept, information and communication technology including industrial internet solutions should be used to optimally integrate the mineral processing chain in the form of distributed network of advanced process controllers and plant mineralogical data.

SUMMARY OF THE INVENTION

[0002] The invention is defined by the features of the independent claims. Some specific embodiments are defined in the dependent claims.

[0003] According to a first aspect of the present invention, there is provided an apparatus comprising at least one processing core; at least one memory including computer program code, the at least one memory and the computer program configured to, with the at least one processing core, cause the apparatus at least to participate in the control of at least one device connected to a furnace, said furnace comprising a plurality of consecutive sections and material flows through the furnace, wherein data is measured from at least one of the sections and the control comprises at least a basic control level and an optimizing control level.

[0004] According to a second aspect of the present invention, there is provided an apparatus in accordance with the first aspect, wherein the control is based at least partly on a reaction rate calculated from at least one section of the furnace. [0005] According to a third aspect of the present invention, there is provided an apparatus in accordance with the first or second aspect, wherein the control is based at least partly on an energy balance calculated from at least one section of the furnace.

[0006] According to a fourth aspect of the present invention, there is provided a method comprising: measuring data from a sample of material; measuring data from a multi-hearth furnace, performing a first determination based at least partly on the measured data, and feedforward controlling a device connected to the furnace based at least partly on the first determination.

[0007] According to a fifth aspect of the present invention, there is provided a method in accordance with the fourth aspect, wherein the control is based at least partly on a reaction rate calculated from at least one section of the furnace.

[0008] According to a sixth aspect of the present invention, there is provided a method in accordance with the fourth or fifth aspect, wherein the control is based at least partly on an energy balance calculated from at least one section of the furnace.

[0009] According to a seventh aspect of the present invention, there is provided an apparatus comprising: at least one processing core; at least memory including computer program code, the at least one memory and the computer program configured to, with the at least one processing core, cause the apparatus at least to participate in controlling at least one device connected to a furnace comprising a plurality of consecutive sections and material flows through the furnace, wherein data is measured from at least one of the sections and the control includes a model predictive control.

[0010] According to an eight aspect of the present invention, there is provided a method comprising: measuring data from a sample of material; measuring data from a furnace, performing a first determination based at least partly on the measured data, and controlling a device connected to the furnace based at least partly on the first determination, wherein the control includes a model predictive control component. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] FIGURE 1 is a flow graph of a mineral calcination process in accordance with at least some embodiments of the present invention;

[0012] FIGURE 2 illustrates a rotating kiln;

[0013] FIGURE 3 illustrates a cross-sectional diagram of a multiple hearth furnace (100), comprising the following elements: feed in (101), in hearth (102), burners (103) and (104), product out (105), out hearth (106), rabble arm (107), gas exhaust (108), exemplary section (109) and automation system (200);

[0014] FIGURE 4 is a graph illustrating the differential scanning calorimetry (DSC) and thermo gravimetric (TGA) curves of kaolin;

[0015] FIGURE 5 is a graph illustrating an exemplary steady state temperature profile for a furnace, wherein the x-axis presents the volumes of the solid and the gas phases arranged in the order of the solid flow from top to bottom;

[0016] FIGURE 6 is a block diagram of multivariate control;

[0017] FIGURE 7 is a block diagram of temperature control;

[0018] FIGURE 8 is a graph illustrating a comparison of the dynamic and static energy balances when the feed rate changes;

[0019] FIGURE 9 is a graph illustrating process variables related to Hearths 6 to 8;

[0020] FIGURE 10 is a graph illustrating process variables related to Hearths 4 to 6;

[0021] FIGURE 11 is a graph illustrating a comparison of the predicted and measured gas consumption in a furnace;

[0022] FIGURE 12 is a block diagram of the overall control strategy;

[0023] FIGURE 13 is a block diagram of a feed forward control strategy for a multi- hearth furnace;

[0024] FIGURE 14 is block diagram of a feedforward strategy based on the soft- sensor using the energy balance in the early hearths; [0025] FIGURE 15 is a flowchart of the main energy fluxes in and out of a furnace;

[0026] FIGURE 16 illustrates the burners configuration in hearth 4;

[0027] FIGURE 17 is a graph illustrating gas temperature and combustion gas flowrate in Hearth 4;

[0028] FIGURE 18 is a graph illustrating hearth 4 temperature and the combustion gas flows;

[0029] FIGURE 19 is a graph illustrating the effect of the combustion gas flow rate on the gas temperature in Burner 1, Hearth 4.

DESCRIPTION OF EMBODIMENTS

[0030] This disclosure describes an innovative, mineralogy-driven approach to control the final product of a mineral production chain by a control strategy based on the feed type characterization and optimization algorithms based on an integrated data bank.

[0031] Calcination can be defined as a process of heating a substance to high temperatures in air or oxygen, causing a loss of water. For many materials, calcination is an important process having a major effect on the characteristics of the final product. Furnaces, such as rotary kilns, shaft furnaces and multiple hearth furnaces (MHF) are widely used in industry for calcination. Industrial minerals which are commonly calcined are for example kaolin, perlite and diatomaceous earth. The main purpose of the calciner control system is to ensure uniform product quality, whereas maximizing the furnace capacity and improving its energy efficiency is necessary for optimal operation.

[0032] Different applications place specific requirements on the properties of materials. The degree and ease with which material properties can be customized during refining varies with mineralogy. The effect of the ore mineralogy on the product quality is twofold. First, various ore properties (like particle size distribution, structure ordering and some impurities) strongly affect the reaction rates and heats, thereby shifting the temperature profile in the furnace and the final product properties. The second effect of ore mineralogy on the product quality relates to some of the impurities directly affecting the final product characteristics without influencing the operating conditions in the calciner. [0033] As it is difficult to measure the product characteristics and the solid temperature profile in the furnace, existing control systems mostly attempt to maintain constant gas temperature using traditional control implementations, such as proportional- integral-derivative control (PID). This strategy helps to attenuate the variations in the solid phase temperature and the calcination reaction rates through the furnace. However, the variations in the solid phase are not eliminated completely because of the varying ore types and mineralogy, and therefore, existing control systems do not allow achieving the uniform product calcination.

[0034] Since the quality of the calcined product heavily depends on the temperature profile along the furnace, the stable desired temperature is acute for production of optimal quality products. However, controlling the temperature in industrial processing equipment is very challenging due to various factors. In particular, the cross-coupling effect between the variables as well as between the zones (hearths) increases the difficulties in maintaining the temperature profile. Thus, in many cases the desired gas temperature profile cannot be efficiently maintained by controlling temperature measurements independently using conventional proportional-integral (PI) control. In the prior art, the underlying chemical and physical properties have not been sufficiently taken into account when devising control strategies. To conclude, many authors have confirmed that the calcination reactions occurring in the furnace are certainly affected by at least ordering, particle size and the heating rate in materials, for example kaolin.

EMBODIMENTS

[0035] To cope with the effects of varying ore mineralogy, this disclosure introduces a three level control strategy for the calciner, kiln, furnace or hearth 100. The basic level controllers aim to stabilize the gas temperature in the furnace, which requires compensating the interactions between the temperature measurements. The stabilizing controllers manipulate the gas temperature to attenuate/compensate variations in the calcination reactions in the solid phase. The optimizing control level adjusts the calcination reaction rates in the optimal manner according to the ore mineralogy and the selected product specifications. Therefore, the proposed control strategy considers the effect of the ore properties on the final product quality and the stabilizing controllers put special emphasis on transition phase control when ore type and its mineralogy is changing. [0036] In the present disclosure, the furnace or kiln 100 is divided into one or more sections 109, alternatively called volumes, portions or segments. In the context of the disclosure, a 'section' is a construed portion of the furnace. Any physical sections may differ from construed sections in some embodiments. It is also possible that the actual physical sections form the basis for the construed sections. Sections in the context of the disclosure are intended to be used for measurement purposes, control purposes or both measurement and control purposes. For example, the temperature of section 1 may be measured and the measurement value subjected to at least one determination, the result of which is used to control a device located in section 5. Sections may be numbered consecutively according to material flows in the furnace or kiln, for example section 1 will be the first portion of the furnace the feed enters and section 8 a portion of the furnace nearest to the product exit. In some embodiments of the disclosure, material flows through the furnace 100, i.e. the material enters via an entrance 101 and exits the furnace via one or more exits 105. The material may enter and exit through the same physical entrance or, in the alternative, the entrance and exit may be different. In some embodiments, the material will circulate through the furnace via each above-mentioned section in turn; however the material may also recirculate multiple times through the same section. A furnace may comprise multiple hearths, some of which further comprise burners (e.g. 103, 104). Some hearths are so-called "in hearths" 102 and some are so-called "out hearths" 106. Gas may exit the furnace via the gas exhaust 108. Material may be circulated in the hearth using a rabble arm 107, or other methods such as a screw.

[0037] The aim of the overall control strategy is to maximize capacity and minimize use of energy at the same time achieving the quality requirements of the products. This will be done primarily by controlling the temperature profile of the furnace by manipulating the gas flows and feedrate. The starting point for the overall control strategy is to determine the setpoints to the temperature profile controllers of the furnace and to the feedrate of the furnace. These setpoint values are very much based on the feed type characteristics (in this disclosure one of them being Fe203). The setpoint values are stored in the integrated databank and displayed to the operator as a look-up table. The look-up table may be in a matrix format where the x-axis comprises the feedrate and y-axis comprises the specific feedtype characteristics (for example Fe203) and the matrix element values comprise the setpoints to the temperature profile values. After processing, the process operation data is added to these setpoint values in the integrated databank for further data analysis. [0038] In this disclosure, a look-up table with its values is determined offline by clustering the process operation data using modern data analytics methods, for example a self-organizing map(SOM) , confirmed by the steady state mass and energy balance calculations and finally checked by the domain experts. The look-up table can be extended/updated when ore types are changing dramatically, for example when changing the processing ore deposit. The look-up table values may be adjusted, modified or updated manually by human operators or generated automatically by extrapolation or other methods.

[0039] To run the furnace, the operator will select first the product type, the desired values of the process, for example the soluble aluminium content of the product and its threshold values based on the feedtype. Next, the setpoints for the temperature profile controllers and the feedrate are set from the look-up table display.

[0040] In some embodiments of the disclosure, the overall control strategy of the furnace includes three level control methods: Optimizing, stabilizing and basic control methods. The methods are in use simultaneously, but in with different control intervals in the control hierarchy and interacting with each other, i.e. several control strategies are online at the same time (i.e. operating in parallel). For example the feed-type and look-up table-based optimizing control may be adjusted on a daily or hourly basis, while simultaneously the stabilizing control strategy is online and adjusted hourly, and the basic control method is adjusted in accordance with typical industrial standards.

[0041] The aim of the optimizing control strategy is to maximize capacity (for example, on an hourly basis), to improve product quality consistency and to minimize the use of energy (on a minute-by-minute basis) with respect to the setpoints coming from the look-up table and constraints from the process condition measurements (for example temperatures, gas flows) and indicators (for example softsensors for reaction rates, mullite content, energy balances) The optimization control will give the setpoints to the stabilizing level control strategy which runs the process as stable as possible in accordance with these setpoints. Ore properties in the feed to the furnace are, however, varying, strongly affecting the furnace operation also as a disturbance. An advantage of monitoring the process in the furnace via indicators (e.g. a reaction rate) is that the final product properties can be predicted and adjusted with minimal loss of material. Furthermore, adjusting the process based on the material in the furnace instead of the final product results in, at the very least, an increased quality of the product which would have undergone processing during the delay between measurement and adjustment which occurs if the process is adjusted based on measurements of the final product.

[0042] Relating to maximizing the capacity of the process, the progress of the calcination in the furnace may be indirectly followed by calculating the mullite content in the furnace via the dynamic energy balance. The mullite content gives an indication how much spinel is formed. Based on data analysis, it appears that a threshold value such as a specific percentage may be calculated for the maximum allowed spinel content in the product. Therefore, a value higher than the threshold value provides room to increase the feed rate of the process, leading to maximization. As previously stated, one goal of optimizing the process is to achieve consistent high product quality, i.e. low mullite content. Therefore, in this disclosure, data analysis is done via a soft-sensor which to predict the mullite content of the product and adjust the process variables accordingly to maximize the process capacity. The operation of the soft-sensor is described in detail later in this disclosure.

[0043] In the present disclosure, the optimizing control level includes formulating the feed rate, desired temperature profile control and other setpoints or reference values for the calcination process. This is achieved at least partly by measuring values from the feedstock and environment, such as, for example, the iron content of the material. The process may include utilization of pre-defined values stored in the look-up table or another form of databank, or database. In some embodiments, the furnace is operated by human operators, who look up the values. The human operators may operate in conjunction with computerized automatic control programs, or the process may be fully automated. The control is based on measuring and calculating the energy balance of the furnace, i.e. the mullite content (calculated from the energy balance) and the reaction rate (calculated from an energy- level model).

[0044] The optimization control will give the setpoints to the stabilizing level control strategy which run the process around these setpoints as stable as possible. Ore properties in the feed to the furnace are, however, varying, strongly affecting the furnace operation also as a disturbance.

[0045] In the present disclosure, the stabilizing control level comprises observing the operations in the early sections of the furnace and adjusting the operations in the subsequent sections aiming to compensate for the observed fluctuations. The control strategy is shown in Figure 13. In some embodiments, where a furnace is divided into at least three sections, the difference between the currently measured and the 'nominal' temperature in a second section is fed to the feedforward controller, acting as a combination of the transport delay from the first section to a third section and a positive gain. The feedforward controller is adjusting the temperature setpoint in the third section, so that the setpoint is decreased when the exothermic reaction is actively ongoing in the first (and the temperature in the second section is high), and the setpoint is increased in the opposite case. Thus, the feed forward controller aims to compensate the variations observed in the first section and achieve a uniform quality of the final product. Furthermore, in some embodiments, the feedforward controller can be modified to consider the gas flow in the first section in addition to the temperature measured in the second section.

[0046] In the present disclosure, feedforward control is used in at least some of the embodiments, in particular with the stabilizing control level mentioned above. Feedforward control means measuring and accounting for disturbances in measured signals pre-emptively. It comprises at least one element or pathway within a control system which passes a controlling signal from a source in its environment, often a command signal from an operator, to a load elsewhere. Feedforward control has proved to be a very powerful process control scheme, especially for disturbance rejection. With feedforward control, compensation for the effect of a measured disturbance is possible before the process is affected. Feedforward controller is thus possible when the disturbance is measured and a process model which includes the effect of disturbance on the output is available. In some embodiments of the disclosure, feedforward control elements may be used in conjunction with open loop, feedback, or both open loop and feedback control elements, the control elements thus comprising the control system. For example, the feed rate of section 1 may be measured and the measurement value subjected to at least one determination, which is then used to adjust the gas flow of a burner situated in section 8.

[0047] Similarly to the previous controller, in another embodiment, the stabilizing control level comprises a control strategy which aims to adjust the temperature in a third section to compensate the fluctuations observed in the earlier sections of the process. In more details, the temperature setpoint for the third section is lowered when the exothermic reaction in the first section is running actively, and the setpoint is raised if the speed of the exothermic reaction in the first section goes down. In order to evaluate the operating conditions in the first section more precisely, a soft-sensor based on the energy balance of multiple sections (occurring early in the hearth) is developed to estimate the intensity of the exothermic reaction in the first section. The soft-sensor inputs include the gas temperature measured at the exit from the furnace and in the second section, the gas flow rates to the first and second section and the solid feed rate. The control strategy is demonstrated in Figure 14. Based on the energy balance calculations, the soft-sensor estimates the rate of the exothermic reaction occurring in the first section. The difference between the estimated and the nominal reaction rate is fed to the feedforward controller, acting as a combination of the transport delay from the first section to the third section and a positive gain. Thus, the temperature in the third section is adjusted according to how the calcination reactions are occurring in the earlier sections, thereby minimizing the final product quality variations.

[0048] In some embodiments of the present disclosure, the data, measurements or control values used in conjunction with the control system include at least one of the following: level of material in a container, volume of material in a container, weight of material in a container, speed of a conveyor, pressure in a container, temperature, feed rate, motor speed, motor torque, motor revolutions, motor voltage, motor current, loss-in-weight data, gas temperature, gas flow, air flow, combustion air, fan speed, air delivery pressure, differential pressure, variables related to oil pumps e.g. voltage or current.

[0049] In some embodiments of the present disclosure, the devices or actuators being controlled include at least one of the following: burners, conveyors including screw conveyors, rotating arms, scrubbers, hatches, doorways, pumps, fans.

[0050] In some embodiments of the present disclosure, soft-sensors are used as part of the control system. Soft sensors or virtual sensors are a common name for software where several measurements are processed together. Commonly soft sensors are based on control theory and in some cases are called state observers. There may be dozens or even hundreds of measurements incorporated into a soft sensor. The interaction of the signals can be used for calculating new quantities that need not be measured. Soft sensors are especially useful in data fusion, where measurements of different characteristics and dynamics are combined. It can be used for fault diagnosis as well as control applications. [0051] An exothermic reaction is a chemical or physical reaction that releases heat. It gives net energy to its surroundings. That is, the energy needed to initiate the reaction is less than the energy released. When the medium in which the reaction is taking place gains heat, the reaction is exothermic. When using a calorimeter, the total amount of heat that flows into (or through) the calorimeter is the negative of the net change in energy of the system.

[0052] Energy is conserved; the rate of change of energy within the control volume equals the net rate of energy transfer into the control volume. The streams entering and exiting a control volume have energy within them, in various energy forms such as internal, potential, and kinetic forms, and all these energy changes sum up to result in energy changes within a system . Coulson, expressed an energy balance based on the first law of thermodynamics to show the energy changes as:

Energy out = Energy in + generation - consumption - accumulation

[0053] The objective of calculating the energy balances is to obtain precise values of the measurements, with the help of laws of conservation of mass and energy. Direct process measurements are very limited as certain measurements cannot be obtained using instruments, therefore, in order to compute missing measurements it necessary to figure out the energy balances.

[0054] The basic control level comprises proportional, integral, and derivative controllers and combinations thereof to control devices connected to the furnace or kiln. These controllers may be, for example, PI or PID controllers. Typically these controllers will have a setpoint relating to a process value and they will adjust the input of the device in order to obtain that value. The type of value and measurement can be e.g. temperature, feed rate, gas flow or any of the data mentioned above in this disclosure.

[0055] Suitable furnaces for calcination in the present disclosure include, inter alia, multiple hearth furnaces, shaft furnaces, rotary calciners and rotary kilns.

[0056] Suitable materials for calcination in the present disclosure include, inter alia, kaolin, perlite and diatomaceous earth.

[0057] In some embodiments of the present disclosure, the control strategy utilizes at least partly model predictive control (MPC). MPC is an advanced control method widely used in various industries for the control of multivariable systems. Model predictive control is based on a dynamic model of the process. MPC may predict the changes in the dependent variables of the modeled system that will be caused by changes in the independent variables. MPC has the ability to anticipate future events and can take control actions accordingly.

[0058] In one embodiment of the present disclosure every hearth of a MHF calciner is equipped with at least one temperature measurement, whereas only hearths 4 and 6 have burners as the actuating elements. Therefore, the gas temperature is maintained as constant in the hearths 4 and 6, while the variations in the solid phase temperature profile due to changing ore types can be observed through the temperature measured in other hearths. Note that the operating conditions in the furnace may vary, disregarding the gas temperature control. Therefore there is a need for a stabilizing control minimizing the final product quality variations via manipulating the setpoints of the basic temperature controllers.

[0059] In some of the embodiments of the present disclosure, the calcination is performed in a multiple-hearth furnace as shown in Figure 3, comprising eight hearths and having counter-current solid and gas flows. The furnace walls are constructed of bricks and are enclosed by a cylindrical steel shell with refractory lining. The heat required for calcination is supplied to the furnace through four tangentially aligned methane burners located on the Hearths 4 and 6. The temperature in the 'fired hearths' is controlled by varying the fuel gas flow, which determines the amount of combustion air. The material flow through the furnace is stirred spirally and moved across the hearths by centrally located vertical rotating shaft carrying arms with rabble blades. Four arms are used on each hearth, and each arm carries three to five rabble blades, whilst material is fed into the top hearth through a single inlet from the weigh feed hopper to the periphery of the hearth. On the odd numbered hearths (i.e. the in hearths), the material is stirred by the rabble blades towards the center of the hearth, and the material drops down to the next hearth from the center through a single annulus around the shaft. In contrast, the material on the even numbered hearths (i.e. the out hearths) is moved outwards to be dropped through the drop holes at the periphery of the hearth to the following hearth. The stirring pattern is repeated until the lowest hearth is reached, from which the calcined product is extracted through the two exit holes. [0060] In an illustrative experiment, a period of time was considered during which the furnace was operated with a feed rate of 100 kg/min. During the considered period, the temperature in the Hearth 6 is tightly controlled, whereas the amount of methane required by this hearth rapidly rises after sample 500 while the temperature in the Hearth 8 drops simultaneously, as it can be seen in Figure 9. In fact, the temperature in Hearth 8 strongly correlates with the temperature of the solids leaving the furnace. Furthermore, the observed drop of the gas and solids temperature in the Hearth 8 is probably caused by lower temperature of the solids leaving from the Hearth 6. At the same time, the methane inflow to the Hearth 4 drops, while the temperature in this hearth is stable and the temperature in the Hearth 5 is even increasing, see Figure 10.

[0061] The observed process variable trends could be explained by partly shifting the exothermic reaction from the Hearth 6 to earlier hearths. This leads to the increase of the temperature in the Hearth 4, which is mostly compensated by the lowered gas flow to the hearth, and also in the Hearth 5. At the same time, less spinel phase is formed in the Hearth 6 causing simultaneous drop of the solid peak temperature reached in this hearth. The effect on the gas temperature in the hearth is fully compensated by the increased methane inflow, but the solid and gas temperature in the subsequent hearths drops. Better understanding of the described phenomena requires more accurate modelling of the heat exchange between the solids and the gas phase, especially in the Hearths 4 to 6.

[0062] A conducted case study confirms that tracking the temperature in the Hearths 4 and 6 is not able to eliminate variations of the temperature profile in the furnace completely. In fact, many similar cases are found in the process data, especially for the high feed rates.

[0063] In brief, the increased reaction rate can be recognized when the gas flow to hearth 4 drops while the gas temperature in hearth 5 rises. The effect propagates to the furnace increasing the temperature in Hearth 1, raising the gas consumption in Hearth 6 and lowering the temperature in Hearths 7 and 8. In order to collect more representative statistics regarding the phenomena, the following indicator r of the exothermic reaction intensity in Hearth 4 has been introduced: r = T ₅ — 2.5 * ₄ , where T ₅ is the temperature measured in Hearth 5 and F ₄ is the gas flow to Hearth 4. [0064] Based on the case study, the indicator is expected to be related to the temperature profile and the gas consumption in the furnace. In order to confirm this, the following linear models describing the total fuel consumption in the furnace have been constructed for different solid feed rates:

F _g = a ₀ (F) + Oi( )r + ₂ (F)r ₆ where F _g is the total fuel consumption in the furnace and the coefficients a ₀ , % and a ₂ depending on the solid feed rate. As all the coefficients % for different feed rates are all negative with the mean value of -0.28, it can be concluded that the high intensity of the exothermic reaction in Hearth 4 decreases the fuel consumption in the furnace.

[0065] The results of the model for the feed rate of 120 kg/min are shown below in

Figure 11 to confirm the model accuracy. Thus, the model proves that the exothermic reaction rate in Hearth 4 is a key factor determining the process conditions in the calciner (together with the temperature in Hearth 6).

[0066] The interaction between the burners may cause a nonlinear correlation between gas flow and temperature in hearth 4. The scenario of burner interactions might be the reason for the rise of nonlinearities during temperature control of a furnace. From Figure 19 it is noticeable for B l, that after an increase of gas flow over 60 m3/h the effect on temperature is negative. This means that a rise in gas flow after this point will cause a decrease in temperature. This effect is similar for the other burners, therefore this phenomena is treated in the same manner for the four burners in hearth 4. Note that nonlinearity reduces the utility of traditional control systems for controlling the system.

[0067] In one or more embodiments, the process control system comprises the optimizing, stabilizing and the basic levels, as shown in Figure 12. The optimizing level aims to select the best operating conditions improving the production capacity, the energy efficiency and achieving the required product quality. The plant personnel determines the product grade taking into account the current ore mineralogy, the client orders and the storage levels. The product quality requirements, for example the soluble aluminium contents and the brightness, are defined according to the selected product grade. The lookup tables provide the setpoints for the gas temperature in Hearths 4 and 6 depending on the current production capacity and iron contents in the ore. The process operators further adjust the temperature setpoints on a regular basis (for example, once a day) based on the laboratory measurements of the product characteristics, aiming to maintain the product quality within the specifications. If the production rate maximization becomes important, the optimization problem is resolved to advise to the operators regarding the possibility to increase the feed rate. Note that the optimizing level may receive input for the setpoints from the system, for example in figure 12 the quantity of soluble aluminum is measured from the furnace and used to adjust the setpoints.

[0068] The stabilizing level aims to attenuate the variation in the calcination reactions taking place in the solid phase of the furnace. In other words, the gas temperature setpoints have to be modified according to how the calcination goes in the solid phase of the furnace. Thus, if the exothermic reaction starts in the earlier volumes in Hearth 6 or occurs actively in Hearth 4, the temperature setpoints need to be lowered to save fuel and avoid over-calcination. In opposite, if the exothermic reaction deactivates or starts later in the furnace, it has to be compensated by rising the setpoints for the gas phase temperature. In order to assess the calcination progress, the energy balances are calculated for the whole furnace and also for its part (first four hearths). Next, the soft sensors (SS2) are developed to estimate the amount of mullite at the final product (using the energy balance for the whole furnace) and to estimate the exothermic reaction rate in hearth 4 (using the energy balance for the first four hearths).

[0069] The overall strategy assumes that the basic temperature controllers in hearths 4 and 6 are operating properly, which means the temperature measured in the Hearths 4 and 6 is close to the provided setpoints. Four burners are installed in each of Hearths 4 and 6, providing energy to four quarters of the Hearths independently. The basic PI controllers are not able to maintain the desired temperature in all quarters of Hearth 4, complicating the implementation of the overall control strategy. Thus, the basic controllers aim to achieve similar conditions in different sectors of Hearths 4.

[0070] An estimation of the solid temperature when leaving hearth 4 is presented on Figure 8 (the x and o lines present the results based on the dynamic and static energy balances respectively). In a first case study, there was a change in the feed rate around samples 150 and 600, whereas in a second case study the feed rate changed around samples 200 and 500. It is clear that the estimations based on the static balance show outliers around the steps in the feed rate, and therefore do not provide satisfactory performance. On the other hand, the outcomes of the dynamic energy balance provide reasonable results in all cases.

[0071] In an illustrative embodiment, the material being processed is kaolin. Kaolin is an important industrial mineral in many different products. Different applications place specific requirements on the properties of kaolin. The degree and ease with which kaolin properties can be customized during refining varies with mineralogy. The effect of the ore mineralogy on the product quality is twofold. First, various ore properties (like particle size distribution, structure ordering and some impurities) strongly affect the reaction rates and heats, thereby shifting the temperature profile in the furnace and the final product properties.

[0072] The second effect of ore mineralogy on the product quality relates to some of the impurities directly affecting the final product characteristics without influencing the operating conditions in the calciner. In particular, the impurities containing iron are known to have a strong impact on the product color. If the iron content is too low, the energy balance and the temperature profiles in the furnace are disturbed. Hence, the in- situ distribution of different types of kaolin and their processing conditions need to be matched up with demand in an optimal manner.

[0073] The route of kaolin from the pit to the calciner can be divided into three primary sections: pit operations, refining processes and drying processes. Pit operations generally involve the breaking down of kaolinized granite forming a suspension of clay and sand, which are thereafter separated. At the refining stage, sand, mica and other impurities contained in the clay suspension are extracted to get a pure clay. In particular, decreasing the iron content improves the brightness characteristics of the final product. Several drying methods, including thickening, mechanical drying and heating, are used to remove moisture from the clay.

[0074] Calcination is one of the most important ways of enhancing the properties and value of kaolin. As a result of calcination, the kaolin becomes whiter and more chemically inert, allowing it to be used in a wide variety of products, such as paper, rubber, paint and refractory items. The calcination process is usually performed in a rotating kiln, as shown in Figure 2, or in a multiple hearth furnace (MHF) as shown in Figure 3. [0075] Kaolin consists primarily of the mineral kaolinite, which has the formula Al ₂ Si205(OH) ₄ . During calcination, kaolin undergoes four physical-chemical processes as listed below.

[0076] First, the evaporation of the free moisture occurs (T<100 °C).

H_2 0(1)→ H_2 0(g)

Next, kaolin undergoes a dehydroxylation reaction, in which the chemically bound water is removed and amorphous metakaolin is formed at 450 - 700 °C.

Al_2 03 ^■ ISiO l ^■ 2H_2 O→ Al_2 0_3 ^■ 2SiO_2 + 2H 2 0(g)

[0077] The third physical-chemical process involves a reaction leading to the transformation of metakaolin to the 'spinel phase' by exothermic re-crystallization at 925- 1050 °C.

2(Al_2 0_3 ^■ 2SiO_2 )→ 2A1 2 0_3 ^■ 3SiO_2 + SiO_2 (amorphous)

[0078] In the fourth and final process, the nucleation of the spinel phase occurs and the material transforms into mullite at temperatures above 1050 °C.

3(2Al_2 0 3 ^■ 3SW 2 )→ 2(3 Al 2 0 3 ^■ 2SW 2 ) + SSiO_2

[0079] Mullite is hard and abrasive, and as a result it can cause damage to process equipment. The desired final consistent product which is within the specification limits has both a low mullite and metakaolin content. The differential scanning calorimetry (DSC) and thermo gravimetric (TGA) curves presenting kaolin calcination are given in Figure 4.

[0080] As stated above, the impurities present in kaolin can be roughly divided into two categories: iron and organic components. The iron components make a very significant difference when calcining kaolin compared to the organic components. Heating the clay matter to high temperatures causes the iron oxides to be oxidised from the green/blue Fe2+ to the red Fe3+ giving the product a shade of pink. Good correlations have been found between the measured iron content and properties of product, like brightness, yellowness and light absorption coefficients, to enable prediction of the colour of the product. In previous studies it has been found out that kaolin having high iron content before beneficiation, but low after refining, is favourably comparable, in terms of brightness, with kaolin that had naturally low iron content once both samples had been calcined. The iron can be removed from clay by several methods, including magnetic separation and reductive chemical bleaching. Removal of iron gives the kaolin a significant added value, but in many cases the separation is not cost effective. Noticeable colour enhancement can also be given to kaolin, when calcining in a reductive atmosphere using gases such as carbon monoxide. In addition, this method can lower the abrasiveness of the product.

[0081] The second category, organic material, can contaminate kaolin by natural and artificial means. Natural contaminants are mainly particles of wood, leaf matter and spores. Artificial contaminants refer to contamination by processing, for example, during the process route from a pit to refineries dispersants, such as polyacrylate, are added to prevent settling. All organic components present in low temperature calcination (-700 °C) have a charring effect giving kaolin grey colour. However, when heating to a higher temperature, such as 1000 °C, it has been generally determined that all organics are removed from the kaolin so that they do not have any effect on the final product.

[0082] Different types of kaolin have to be treated differently in the furnace, according to the application requirements. In particular, the iron contents in the ore, having a strong effect on the product brightness, have to be considered while selecting the optimal operating conditions. In addition, the solid feed rate and its residence time distribution have a major impact on the temperature profiles in the furnace. Therefore, both the solid feed rate and the iron contents of the ore are considered to define the target temperature profile in the furnace. Setpoints for the hearth 4 and hearth 6 temperatures according to the feed rate and the iron contents can be defined as process know-how based look up variables.

[0083] Alternatively, the setpoints for hearths 4 and 6 can be selected by resolving an optimization problem aiming to maximize the calciner capacity/ minimize the energy consumption while maintaining the acceptable product quality. As an example, the optimization can be mathematically formulated as follows:

max F I rain F ₄ + F ₆

with respect to the constraints:

F ^T F ₄ , F ₆ , r, F) = T ₄ , F ^T6 (F ₄ , F ₆ , r, F) = T ₆ ,

S(F ₄ , F ₆ , r, F)≤S ^* ,

•γΊηίη < T < imax where F, F ₄ and F ₆ represent the feed rate, the gas flows to hearths 4 and 6, r the current value of the indicator introduced in (5), T ₄ and T ₆ are the temperature in hearths 4 and 6, S and S ^* are the soluble aluminium and/or mullite contents and their thresholds (if applicable).

[0084] Based on the Applicant's understanding of the prior art, maintaining a constant gas temperature profile in a furnace does not usually stabilize the conditions in the solid phase. In the considered process, the variations in the solid phase are caused by fluctuating ore properties, which can affect the reaction rates, reaction heats and other important process parameters. The variations are especially visible around the hearth 4, where the exothermic reaction running in the solid phase is changing its intensity in an unpredictable way. The temperature in the hearth 4 is controlled, and therefore, it does not reflect the operating conditions in the hearth, as confirmed by the example provided. Instead, the temperature measured in hearth 5 and the gas flow to hearth 4 are both very sensitive to the exothermic reaction speed in the hearth 4.

[0085] Thus, in some of the embodiments described herein, the control strategy is based on observing the operations in the early hearths of the furnace (hearths 1-4) and adjusting the operations in the subsequent hearth 6 aiming to compensate the observed fluctuations. This embodiment could be described as a feedforward control strategy to attenuate variations of the calcination reactions. The control strategy is illustrated in Figure 13.

[0086] The difference between the currently measured and the 'nominal' temperature in the hearth 5 is fed to the feedforward controller, acting as a combination of the transport delay from the hearth 4 to the hearth 6 and a positive gain. The feedforward controller is adjusting the temperature setpoint in the hearth 6, so that the setpoint is decreased when the exothermic reaction is actively ongoing in the hearth 4 (and the temperature in the hearth 5 is high), and the setpoint is increased in the opposite case. Thus, the feed forward controller aims to compensate the variations observed in hearth the 4 and achieve a uniform quality of the final product. Furthermore, the feedforward controller can be modified to consider the gas flow in hearth 4 in addition to the temperature measured in the hearth 5.

[0087] Similarly to the previous embodiment, this control strategy aims to adjust the temperature in the hearth 6 to compensate the fluctuations observed in the earlier hearths of the process. In more details, the temperature setpoint for the hearth 6 is lowered when the exothermic reaction in hearth 4 is running actively, and the setpoint is raised if the speed of the exothermic reaction in the hearth 4 goes down. In order to evaluate the operating conditions in the hearth 4 more precisely, a soft- sensor based on the energy balance of first four hearths is developed to estimate the intensity of the exothermic reaction in the hearth 4. The soft-sensor inputs include the gas temperature measured at the exit from the furnace and in hearth 5, the gas flow rates to the hearths 4 and 6 and the solid feed rate. The control strategy is demonstrated in Figure 14. Based on the energy balance calculations, the soft- sensor estimates the rate of the exothermic reaction occurring in the hearth 4. The difference between the estimated and the nominal reaction rate is fed to the feedforward controller, acting as a combination of the transport delay from the hearth 4 to hearth 6 and a positive gain. Thus, the temperature in hearth 6 is adjusted according to how the calcination reactions are occurring in the earlier hearths, thereby minimizing the final product quality variations.

[0088] In one illustrative embodiment, a model predictive control (MPC) can be applied to the temperature profile control in the furnace. MPC is a multivariate control technology able to handle process constraints and to incorporate economic objectives. For the considered MHF, the MPC manipulates the temperature setpoints in the hearths 4 and 6, while maintaining the product quality above the defined limit while minimizing the total fuel consumption of the furnace. However, the MPC implementation requires a state-space process model, whereas the first principal models, such as the one developed by Eskelinen et al., are not suitable.

[0089] Based on the detailed model described in Eskelinen et al. (2015), a simplified model is developed in the nonlinear Hammerstein- Wiener form, which defines a specific type of nonlinear state space models suitable for example for Model Predictive Control (MPC) implementation. The simplified model aims to preserve the key physical-chemical phenomena taking place in the furnace and to reproduce the nonlinear dependencies between the input and output variables. A simplification of the mechanistic model is designed, describing the dynamics and the nonlinear behavior of the system separately. In more details, the simplified model is expressed as a Hammerstein-Wiener model (HWM), decomposing the model in blocks containing the nonlinearities in static form and the linear dynamics. The linear block, enclosing the dynamics of the process, is preceded and followed by a static non-linear blocks.

[0090] The dynamics of the MHF includes the very fast component related to the gas phase, and the slower component representing the solid state. For MPC implementation, the temperature of the solid has to be described dynamically. Furthermore, as the temperature of the inner layer of the walls has a direct effect on the solid-walls heat exchange; it is also considered as a model state. The simplified model is implemented as follows: x _t+1 = ax _t + (1 - a) ( _t )

y _t = G (u _t , Xt where u _t is a vector containing the process inputs (kaolin feed rate, gas flows to the Hearths 4 and 6), x _t is the state vector contains the temperature of the solids in each volume of the furnace and the internal wall temperature in the hearths, y _t contains the gas phase temperature next to the walls in the hearths, is the time constant parameter of the linear dynamic part of the modeling equations. The time constant is obtained for each modeling equation by identification performed using the MATLAB® identification toolbox. F(u _t ) is a static nonlinear function calculating the steady state of the furnace using the process input values. In order to implement the first function F(u _t ), a look up table has been created by running the mechanistic model simulations with different process inputs. The obtained values are interpolated as follows: f{u _t ) b _{i ik} hf (F _K )hJ(F _g ,)h _k ^z (F _g6 )

[0091] where b _{£ <fe} are the values from the look-up table and the piecewise linear basis functions hf, h and have been used for the interpolation.

[0092] The second function G (u _t , x _t ), involved in the modeling equations (8), calculates the gas temperature profile next to the walls in the Hearths based on the current furnace state x _t and the process inputs u _t . The function G (u _t , x _t ) is implemented by solving the energy balance for the gas phase derived from the mechanistic model.

[0093] The exothermic reaction rate in Hearth 4 can be added to the model (8) as a disturbance and estimated on-line by the means of the Kalman filter. There is a choice how to specify the objective and constraints for the MPC. In particular, as the solid temperature profile is included to the model state x _t , it is possible to directly control the solid phase temperature. Furthermore, the feed rate, considered as one of the inputs of model (8), can be used as a manipulated variable of the MPC, which creates a basis for incorporating the capacity maximization to the MPC objectives.

[0094] Some embodiments may make use of one or more soft sensors, as defined earlier in the disclosure. An illustrative description of the soft sensors follows. A first soft sensor is based on the energy balance of the furnace, as presented in Figure 15. The energy is supplied to the furnace with the inlet flows, including the solid inflow, the draft and cooling air, the fuel and combustion air flows. However, the most of energy is provided by the fuel combustion occurring in the furnace. The enthalpy of the gas, solid and cooling air outflows is estimated and summed up with the energy losses caused by the convection through the walls. Furthermore, the calculations take into account the reactions and the water transferred from the solid to the gaseous phase while computing the amount and the composition of the outflows. In addition, the energy consumed by the first three reactions (water evaporation, dehydroxiliation and formation of the spinel phase) is computed assuming that these reactions are completed by the end of the process. The outcome of the first soft- sensor is the difference between the energy provided to the process and the enthalpy of the outflows and the heat consumed by the reactions.

[0095] The soft- sensor calculations require estimating the amount of solid in the hearths, which can be achieved by using the solid distribution between the hearths obtained from pilot experiments. In the simplest case, these calculations are based on the current feed rate, which results in the static soft- sensor providing misleading results during transients. Due to the long residence time of the solid phase in the furnace, the dynamic soft sensor utilize the past values of the feed rate to calculate the amount of solid in the hearths, thereby remarkably improving the soft-sensor performance during transitions. [0096] A second soft-sensor aims to provide an estimation of the mullite contents in the final product based on the outcome of the first soft-sensor. If the kaolin spends too long in the furnace it is over-calcined and it forms an abrasive mullite structure which can damage the process equipment. Thus, the calcination reaction should be controlled to avoid significant mullite contents in the final product. As mullite is formed by the endothermic reaction occurring at high temperature, it is possible to estimate its amount according to the following equation:

[0097] where m _m is the production rate of mullite, E _x is the output of the first soft sensor (the energy balance of the furnace), and H _m is the energy absorbed while forming 1 kg of mullite. As previously stated, the estimation of mullite contents is a key part of maximizing the capacity of the process.

[0098] The configuration of the burners inside the furnace gives a clear idea of the interactions that occur inside the furnace. In Figure 16 it is showed the burner configuration. This configuration is disposed in four equidistant points around the circumference of the hearth. Every burner is located in one of the points and they are facing a contiguous burner as represented in the figure above. Burner 1 (B l) faces burner 4 (B4), B4 faces Burner 3 (B3) and so on.

[0099] In Figure 17 real data of the process is depicted, demonstrating the behavior of H4. In general one burner temperature, at a given time, follows the setpoint. In the right figure the gas flows for the burners are shown. In this case only one gas flow is manipulated while the rest remain in saturation.

[00100] The top graph in Figure 18 shows the temperature of every burner, from Bl to B4. The bottom image displays the gas flow of the burners, using the same color relations. At first glance, it is perceptible the relation between temperature of Bl and gas flow of B2. This relation is direct, indicating that an increase of gas flow in B2 causes an increase of temperature in Bl, this occurs most probably due to the burner configuration and the fact that B2 faces directly to Bl. A second observation provides that there is a small effect on temperature of B2, even when gas flow of B2 is fluctuating. After observing closely, the gas flow in B2 has an approximate average of 60 m3/h. [00101] The interaction between the burners may cause a nonlinear correlation between gas flow and temperature in hearth 4. The scenario of burner interactions might be the reason for the rise of nonlinearities during temperature control of a furnace. From Figure 19 it is noticeable for Bl, that after an increase of gas flow over 60 m3/h the effect on temperature is negative, this means that a rise of gas flow after this point will cause a decrease in temperature. It is important to emphasize that this effect is similar for the other burners, therefore this phenomena is treated in the same manner for the four burners in hearth 4. Again, this emphasizes the need for more intelligent control systems for controlling the calcination process.

[00102] In some embodiments, at least part of the overall control strategy will be implemented by a plant automation system(200), i.e. computing hardware connected to the furnace. In some embodiments, a programmable logic controller, PLC, may be used as a controller of some or part of the system. The standard IEC 61131 defines programmable logic controllers, however, in the present disclosure, the term PLC is not limited to the standard. The PLC may be connected to the devices by a variety of means, including but not limited to electrical wires. Another type of controller may comprise, for example, a computing device such a server, node or cloud computing device. Comprised in the computing device is a processor, which may comprise, for example, a single- or multi-core processor wherein a single-core processor comprises one processing core and a multi-core processor comprises more than one processing core. The processor may comprise more than one processor. A processing core may comprise, for example, a Cortex-A8 processing core by ARM Holdings or a Steamroller processing core produced by Advanced Micro Devices Corporation. The processor may comprise at least one Qualcomm Snapdragon and/or Intel Core processor, for example. The processor may comprise at least one application-specific integrated circuit, ASIC. The processor may comprise at least one field-programmable gate array, FPGA. The processor may be a means for performing method steps in the computing device. The processor may be configured, at least in part by computer instructions, to perform actions.

[00103] In some embodiments, a network is used to facilitate communication to and from controllers, devices and other elements. The network technologies comprise: wireless local area network, WLAN, Ethernet, universal serial bus, USB, and/or worldwide interoperability for microwave access, WiMAX, standards, and satellite communication methods, for example. Alternatively or additionally, a proprietary communication framework may be utilized. In some embodiments, separate networks may be used for one or more of the following purposes: communication between controllers, communication between controllers and devices, communication between controllers and servers, et cetera.

[00104] As stated above, in relation to the current disclosure, the status of the system, a controller or a device may be adjusted according to various determination criteria comprising: preprogrammed instructions, communications received from other devices including the presence or absence of first or third party devices, elapsed time, preset time, measurements internal or external to the calcination process, user inputs and timing, battery levels, network data, detected usage, planned usage or any combination thereof. Controllers may also have preprogrammed activities or processes. These may be triggered based on the criteria listed above.

[00105] It is to be understood that the embodiments of the invention disclosed are not limited to the particular structures, process steps, or materials disclosed herein, but are extended to equivalents thereof as would be recognized by those ordinarily skilled in the relevant arts. It should also be understood that terminology employed herein is used for the purpose of describing particular embodiments only and is not intended to be limiting.

[00106] Reference throughout this specification to one embodiment or an embodiment means that a particular feature, structure, or characteristic described in connection with the embodiment is included in at least one embodiment of the present invention. Thus, appearances of the phrases "in one embodiment" or "in an embodiment" in various places throughout this specification are not necessarily all referring to the same embodiment. Where reference is made to a numerical value using a term such as, for example, about or substantially, the exact numerical value is also disclosed.

[00107] As used herein, a plurality of items, structural elements, compositional elements, and/or materials may be presented in a common list for convenience. However, these lists should be construed as though each member of the list is individually identified as a separate and unique member. Thus, no individual member of such list should be construed as a de facto equivalent of any other member of the same list solely based on their presentation in a common group without indications to the contrary. In addition, various embodiments and example of the present invention may be referred to herein along with alternatives for the various components thereof. It is understood that such embodiments, examples, and alternatives are not to be construed as de facto equivalents of one another, but are to be considered as separate and autonomous representations of the present invention.

[00108] Furthermore, the described features, structures, or characteristics may be combined in any suitable manner in one or more embodiments. In this description, numerous specific details are provided, such as examples of lengths, widths, shapes, etc., to provide a thorough understanding of embodiments of the invention. One skilled in the relevant art will recognize, however, that the invention can be practiced without one or more of the specific details, or with other methods, components, materials, etc. In other instances, well-known structures, materials, or operations are not shown or described in detail to avoid obscuring aspects of the invention.

[00109] While the forgoing examples are illustrative of the principles of the present invention in one or more particular applications, it will be apparent to those of ordinary skill in the art that numerous modifications in form, usage and details of implementation can be made without the exercise of inventive faculty, and without departing from the principles and concepts of the invention. Accordingly, it is not intended that the invention be limited, except as by the claims set forth below.

[00110] The verbs "to comprise" and "to include" are used in this document as open limitations that neither exclude nor require the existence of also un-recited features. The features recited in depending claims are mutually freely combinable unless otherwise explicitly stated. Furthermore, it is to be understood that the use of "a" or "an", that is, a singular form, throughout this document does not exclude a plurality.

[00111] The work leading to this invention has received funding from the European Union Seventh Framework Programme FP7/2007-2013 under grant agreement n° 310645.

INDUSTRIAL APPLICABILITY

[00112] At least some embodiments of the present invention find industrial application in controlling calcination furnaces.

ACRONYMS LIST

DSC Differential Scanning Calorimetry

FOPDT First order plus dead-time model MHF Multiple Hearth Furnace

MPC Model Predictive control

PI Proportional-integral control

PID Proportional-integral-derivative control

SOM Self-Organizing Map

TG Thermo-gravimetric

REFERENCE SIGNS LIST

CrrATION LIST

NON PATENT LITERATURE

Eskelinen, A. Dynamic modelling of a multiple hearth furnace, 2014, Espoo: Aaltodoc, Aalto University

Eskelinen A., Zakharov A., Jamsa-Jounela S.-L., Hearle J., Dynamic modelling of a multiple hearth furnace for kaolin calcination. AIChE Journal, 2015; 61:3683-3698.