CRYSTALLIZATION and SEPARATION of Kcl FROM

Available on line at Association of the Chemical Engineers of Serbia AChE

Chemical Industry & Chemical Engineering Quarterly www.ache.org.rs/CICEQ Chem. Ind. Chem. Eng. Q. 24 (3) 239−249 (2018) CI&CEQ

J.A. TAVARES CRYSTALLIZATION AND SEPARATION OF L.F. MOURA KCl FROM CARNALLITE ORE: PROCESS A. BERNARDO DEVELOPMENT, SIMULATION AND M. GIULIETTI ECONOMIC FEASIBILITY Chemical Engineering Department, Federal University of São Carlos Article Highlights (UFSCar), Rodovia Washington • Evaluation of alternatives for producing KCl from carnallite by crystallization

Luiz, São Carlos, SP, Brazil • Process synthesis for KCl production from KCl-MgCl2-H2O system phase diagram with saturated NaCl SCIENTIFIC PAPER • Feasibility evaluation of a KCl crystallization plant considering capital and operating expenses UDC 553.3/.4(81):546.32’131:66 Abstract Given the increasing demand for potassium in Brazil, the mining and use of carnallite is becoming increasingly important, because the current source of potassium, sylvinite, is being depleted and there is a risk of shortages. Based on theoretical and practical data available in existing literature, this work describes the development, simulation, and economic feasibility of a process for dissolution and crystallization of potassium chloride from carnallite ore. Positive results were obtained following the application of the Hoffman diagram and determination of the corresponding equation. The proposed process provided over 85% potassium chloride crystallization, demonstrating its superior performance, compared to existing procedures. Keywords: carnallite, potassium chloride, crystallization, process synthesis, economic assessment.

Potassium is the seventh most abundant ele- Evaporite deposits are therefore the most important ment on Earth, accounting for around 2.4% of its crust sources of potassium salts, because the salts present by mass. It is only found in the form of compounds, in these deposits are soluble in water and can be due to its high reactivity and association with other mined and processed more easily. The main evapo- elements, with chlorides and sulfates being most rite deposits of potassium that can be highlighted, common, at levels above 10% in many minerals. The together with their K2O equivalents, are as follows: highest contents of potassium are found in evaporite sylvinite (KCl, 63% K2O); carnallite (KCl⋅MgCl2⋅6H2O, minerals and potassium silicates. Although silicates 17% K2O); kainite (KCl⋅MgSO4⋅3H2O, 19.3% K2O); contain between 10 and 20% of K2O equivalent and langbeinite (K2SO4⋅2MgSO4, 22.7% K2O); polialite are abundant on Earth, they do not represent import- (K2SO4⋅MgSO4⋅2CaSO4⋅2H2O, 15.6% K2O); schoenite ant potassium resources because they are not sol- (K2SO4⋅2MgSO4⋅4H2O, 25.7% K2O); singernite uble in water and their chemical bonds are difficult to (K2SO4⋅CaSO4⋅H2O, 28% K2O) [2]. break, making their mineralization impossible [1]. The utilization of potassium as a plant fertilizer dates from ancient times. There are references to its use since the 3rd century BC, when it was obtained Correspondence: A. Bernardo, Chemical Engineering Depart- from the ashes of wood and salting from marine salts, ment, Federal University of São Carlos (UFSCar), Rodovia Wash- ington Luiz, km 235, São Carlos, SP, Brazil. but a process for potassium salt production from car- E-mail: [email protected] nallite for use as fertilizer was only proposed by Frank Paper received: 19 January, 2017 [3]. Potassium is a key mineral nutrient for plants and Paper revised: 24 August, 2017 Paper accepted: 28 September, 2017 animals, and is the third most abundant mineral ele- https://doi.org/10.2298/CICEQ170119036T ment in our bodies, only exceeded by calcium and

239 J.A. TAVARES et al.: CRYSTALLIZATION AND SEPARATION OF KCl… Chem. Ind. Chem. Eng. Q. 24 (3) 239−249 (2018) phosphorus. Over 85% of the potassium in the human of carnallite, which are transferred to a decanter flask, body is found in key organs, and neither animals nor where an elutriator separates the solutions of pot- plants can live without an adequate potassium intake. assium chloride and sodium chloride. The sodium Belarus (57%), Canada (17%) and Russia (10.4%) chloride solution is transferred to a separation flask, are globally the three largest producers of potassium. where the sodium chloride is removed, and a portion Due to the small domestic production, compared to of this solution is then recycled to the decanter. The the country’s large demand for potassium, Brazil is a potassium chloride separated in the decanter is rem- major importer of potassium fertilizer, and in 2013 oved by centrifugation. The remaining potassium imported 7.6 million t of potassium chloride (KCl) from chloride solution is transferred to a cooling flask, in its main suppliers: Canada (31.59%), Germany which the temperature is reduced to 115.7 °C. This (16.46%), Russia (15.63%), Belarus (14.86%) and cooled solution is transferred to another separation Israel (9.43%). Globally, over 95% of the potassium flask, where carnallite and bischofite are separated. produced is used as fertilizer, with 90% in the form of The bischofite is stored and the carnallite is recycled potassium chloride, and the remainder is consumed to the fusion vessel. by chemical industries [4]. The US3642454 patent [7] describes the crys- Brazilian mineral reserves containing potassium, tallization of potassium chloride from carnallite. This measured as K2O, are estimated to be 16 billion t, process begins with the dissolution of carnallite, according to the Brazilian National Department of preferably with fresh water in order to be able to dis- Mineral Production (DNPM) [4], including the reserves solve all the magnesium chloride, which ranges from of carnallite and sylvinite. The Brazilian regions in one to two and a half times the weight of carnallite. which there are proved reserves of potassium are After this stage, the resulting solution and the Sergipe and Amazonas states. In Sergipe, potassium undissolved salts of potassium chloride and sodium reserves are estimated in 14.4 billion t of carnallite chloride are separated. The separated solid salts are and 480 million t of sylvinite, with about 8.31 and dissolved in the next stage, preferably using fresh

10.40% of K2O, respectively. According to Warren [5], water, generating a solution containing sodium chlo- the Sergipe deposit is composed of stacked cycles of ride and potassium chloride, which goes through a halite, carnallite and sylvite up to 800 m thick. More process of solar evaporation to crystallize the salts than 750,000 t/year of potash ore is extracted by Vale with a high level of potassium chloride (between 66 in its conventional Taquari-Vassouras potash mine. and 76% of the crystals). Finally, the crystallized salts The deposit is small, estimated to be 11 million t of are sent for refining by a washing process, which

K2O, and the mine has an operating cost of $174 per t increases the purity of the crystals to 95% potassium and a projected productive life of nine years. In Ama- chloride. zonas, mineral reserves containing potassium are Patent US6022080 [8] developed for producing estimated to be 1 billion t of sylvinite containing about potassium chloride (KCl) from carnallite consists in a

18.47% of K2O equivalent. These deposits are in the process in which water is heated in a pond by solar localities of Fazendinha and Arari, in the Nova Olinda energy up to a value between 60 and 80 °C, and is do Norte region, and there is no project for exploiting sequentially introduced into a cavern to solution mine this area [4]. the carnallite layer and produce the brine. This brine The crystallization and separation operations is pumped up to surface and fed in a natural evapor- aimed to obtain the greatest possible amount of the ator coupled to an exhauster (crystallizer) which cools desired salt. In order to achieve this, there are several the solution to a temperature between 30 and 40 °C, processes described in the literature that can be crystallizing sodium chloride (NaCl) and potassium considered. The process of diluting carnallite is the chloride (KCl) as final products. most widely used method, resulting in a solution that Patent US8282898 [9] presents a crystallization is heated to 105 °C in an evaporator and then cooled process for high purity KCl from carnallite. The des- in a crystallizer to produce the potassium chloride cribed steps are the dissolution of carnallite to form a salt. Several patents have been registered for this solution containing NaCl, magnesium chloride and process. KCl. The concentration of MgCl2 must be controlled According to the US4140747 patent [6], the between 12 and 25% weight to avoid co-precipitation production of potassium chloride and magnesium of NaCl. The patent claims that if the NaCl weight chloride from carnallite is achieved by feeding car- concentration is not greater than 2% in solution, KCl nallite into a fusion vessel that is heated to 167.5 °C. crystals will have high purity. This results in a suspension of sylvinite and a solution

240 J.A. TAVARES et al.: CRYSTALLIZATION AND SEPARATION OF KCl… Chem. Ind. Chem. Eng. Q. 24 (3) 239−249 (2018)

Due to increasing demand in Brazil, the exploit- and K2Cl2, were read for each curve (dissolution tem- ation of carnallite is essential in order to supply part of perature) in the original diagram. A linear regression the national demand for this alkali metal, since the was performed to describe each of the curves gen- current main natural source, sylvinite, is becoming erated in Origin and MS Excel. With all the equations depleted, leading to a risk of shortages. obtained for each temperature, another linear regres- The current consumption (production + import- sion was performed, generating a general equation ation–exportation) is around 8.1 million t of KCl, and that could effectively describe the entire diagram. this amount minus the production of 0.49 million t of Simulation of the process KCl results in an internal demand deficit of 7.69 million t of KCl, which is indicative of the potential for The simulation was performed in MS Excel, increased domestic potassium production in Brazil [4]. using the reconstructed diagram for KCl-MgCl2-H2O Therefore, the goal here was the development, saturated with sodium chloride, and its generalized simulation, and economic feasibility evaluation of new regression. In this simulation, the mass and energy processes for crystallization of potassium chloride, balances were performed for all stages and all com- aiming to increase production and generate less ponents of the process, resulting in an outline flow- waste. The approach adopted was to apply technical chart of the process, divided into stages [11,12]. The and experimental data available in the literature to the economic feasibility evaluation was also performed in process of dilution and crystallization of potassium this simulation, with calculation of the capital and chloride from carnallite ore. operational expenditures required to implement the process [11]. EXPERIMENTAL Mass and energy balances The process was based on the dissolution of The balances of mass and energy in each step carnallite by the injection of water into the deposit bed were determined following the NaCl-KCl-MgCl2-H2O by means of a drilling system. The dissolution of car- solubility diagram, considering the temperatures and nallite provides a concentrated solution of potassium individual operations required for the partitioning of chloride (KCl), sodium chloride (NaCl), and magne- components [10]. sium chloride (MgCl2), which is pumped into a tank on Economic feasibility of the process the surface, feeding the process of crystallization and separation of the potassium chloride. The process of The economic feasibility of the process was crystallization may take place in several cycles of investigated based on the solubility diagram for the water evaporation and solution cooling, according to NaCl-KCl-MgCl2-H2O system, considering tempera- the path established using the solubility diagram for ture variations, and was based on the use of two websites that provide cost estimations for chemical KCl-MgCl2-H2O saturated with sodium chloride, considering the temperature [10]. The development of processes: matche.com [13] and mhhe.com [14], this study was divided into reconstruction of the employing information available in the specialist solubility diagram available in the literature [10], literature [15,16]. The costs of labor were also con- simulation of the processes, development of a pro- sidered, taking account of labor laws and the number cess flowchart, mass and energy balances of the of employees, as well as the output of the simulated processes, and evaluation of economic feasibility, all process according to the size of the equipment and following methods available in the literature. the stages of the operation. The economic feasibility study also took into consideration the location factor Solubility diagram (Big Mac Index) [17] and the sector inflation (CEPCI, A reconstruction was performed of the solubility Chemical Engineering Plant Cost Index) [18] in est- diagram developed by Hoffman for the quaternary imation of the costs of construction of the site in Bra- mixture of NaCl-KCl-MgCl2-H2O from carnallite dis- zil, the raw material costs, and the energy required to solved in water with saturated sodium chloride, at dif- supply all the units of the site. Finally, standardized ferent temperatures. Data were obtained from the site costs were included for maintenance and repair, original diagram, in which the saturation concen- supply services, laboratories, royalties, taxes, insur- tration of magnesium chloride in water (mol of MgCl2/ ance, depreciation, administration, distribution and

/1000 mol of H2O) is a function of the saturation con- sales, research and development, and variable costs centration of potassium chloride in water (mol of [15,16].

K2Cl2/1000 mol of H2O). In the procedure used, at least 13 points, with corresponding values of MgCl2

241 J.A. TAVARES et al.: CRYSTALLIZATION AND SEPARATION OF KCl… Chem. Ind. Chem. Eng. Q. 24 (3) 239−249 (2018)

RESULTS AND DISCUSSION errors, compared to below the eutectic point, although the values in this range were nonetheless acceptable Diagram for use. Table 1 presents all the equations obtained in The flowchart developed according to the meth- the regressions, together with their corresponding odology outlined above is shown in Figure 1, com- coefficients, from which a single equation could be pared to the original [10]. The substantial similarity obtained for the entire diagram. Hence, a single between them suggests that the developed diagram equation could be used to represent all the values in presented a minimal margin of error throughout the the Hoffman diagram, limited by the eutectic line for process simulation. the solution and the carnallite and potassium chloride After creating the diagram, the analysis was solids. performed using the part of the diagram necessary for The equations presented in Table 1 were used simulation of the process. This part corresponded to in a new regression to obtain a single equation that all values below the E curve (the equilibrium between represented all of them. The method adopted was to the solid phases of carnallite and potassium chloride) perform the regression for each coefficient, a, b and c, in the diagram, as shown in Figure 1. according to temperature. Hence, in performing the From the flowchart of Figure 1, the reliability of regressions, the dependent variables were the coef- each line was evaluated separately, according to tem- ficients a, b and c, and the independent variable was perature, with the lines represented in the form of the temperature. This regression enabled the deter- equations obtained by least squares polynomial reg- mination of a single value for each coefficient of the ression. equation, according to temperature. The equations The results of the polynomial regressions high- follow below. Eqs. (1)–(3) show the regressions for lighted the limitation imposed by the eutectic line for coefficients a, b and c, respectively, according to tem- the solution and the carnallite and potassium chloride perature (T). solids (curve E), which could be described by poly- aTT=+97.2520 0.1786 + 0.0025 2 (1) nomial equations that differed minimally from the real data of the diagram. It could also be seen that use of bT=−10.53 + 1.3839ln (2) the diagram above the eutectic point resulted in larger

Figure 1. Part of the solubility diagram for the KCl-MgCl2-H2O system with saturated sodium chloride, recreated up to the eutectic curve (E) of the solution with the carnallite and potassium chloride solids. The diagonal line passing through the origin indicates the solution of carnallite with pure water.

242 J.A. TAVARES et al.: CRYSTALLIZATION AND SEPARATION OF KCl… Chem. Ind. Chem. Eng. Q. 24 (3) 239−249 (2018)

Table 1. Equations and correlation coefficients obtained by polynomial regressions for different temperatures. CMgCl2: mol of MgCl2/1000 mol H2O; CK2Cl2: mol of K2Cl2/1000 mol H2O

Temperature, °C Equation (a + bx + cx2) R2 2 10 CMgCl2 = 100.49 – 7.5061CK2Cl2 + 0.0552CK2Cl2 0.9992 2 20 CMgCl2 = 101.42 – 6.0066CK2Cl2 + 0.0454CK2Cl2 0.9990 2 35 CMgCl2 = 105.99 – 5.3859CK2Cl2 + 0.0289CK2Cl2 0.9996 2 55 CMgCl2 = 116.19 – 4.9904CK2Cl2 + 0.0331CK2Cl2 0.9994 2 75 CMgCl2 = 123.75 – 4.3696CK2Cl2 + 0.0238CK2Cl2 0.9994 2 90 CMgCl2 = 134.20 – 4.4297CK2Cl2 + 0.0271CK2Cl2 0.9995 2 105 CMgCl2 = 142.74 – 4.2705CK2Cl2 + 0.0248CK2Cl2 0.9986

cT=−0.0843 0.013ln (3) magnesium chloride calculated using the developed equation, both as a function of the same predefined From the three regressions, it was possible to value for potassium chloride. This comparative obtain a single general equation representing the analysis of the points was performed for temperatures Hoffman diagram [10], as shown in Eq. (4): of 25, 35, 55, 75, 90, and 105 °C, according to the CabCcC=+ + 2 (4) Hoffman chart [10]. MgCl22222 K Cl K Cl

Substituting Eqs. (1)-(3) in Eq. (4) gives Eq. (5): Simulation of the process The theoretical basis for the production of potas- CTT=+97.2520 0.1786 + 0.0025 2 + MgCl2 sium chloride from carnallite is depicted in Figures 2-4 +−()10.53 + 1.3839lnTC + (5) (trajectories A, B and C, respectively) for tempera- KCl22 tures between 25 and 105 °C, according to the sys- +−()0.0843 0.013lnTC2 KCl22 tem developed, which was based on the Hoffman In the simulation, Eq. (5) could then be used to chart [9]. Accordingly, three trajectories for potassium represent any point below the E curve in the diagram. chloride crystallization were developed. Trajectory A The application of this equation is presented in Table (Figure 2) begins with the addition of water to the car- 2, where the relative standard deviation was calcul- nallite well to decompose the mineral and generate ated considering the value for magnesium chloride the salting solution, represented by point 1, which provided by the Hoffman chart [9] and the value for then undergoes evaporation until reaching a tempera-

Table 2. Calculation of relative standard deviations (RSD) between the values obtained using the developed equation and the values obtained from the Hoffman chart7 for temperatures of 25, 35, 55, 75, 90, and 105 °C

Chart Equation Chart Equation Chart Equation RSD / % RSD / % RSD / % K2Cl2 MgCl2 MgCl2 K2Cl2 MgCl2 MgCl2 K2Cl2 MgCl2 MgCl2 25 °C 35 °C 55 °C 13.30 30.00 29.83 0.41 15.50 30.00 28.52 3.59 20.00 30.00 27.35 6.54 12.35 35.00 34.58 0.85 14.35 35.00 33.69 2.70 18.60 35.00 32.64 4.93 11.00 40.00 41.47 2.56 13.20 40.00 38.96 1.86 17.00 40.00 38.84 2.08 9.10 50.00 51.43 1.99 10.80 50.00 50.29 0.40 14.50 50.00 48.84 1.66 8.10 55.00 56.79 2.26 10.00 55.00 54.15 1.10 13.40 55.00 53.37 2.13 7.10 60.00 62.23 2.58 9.00 60.00 59.06 1.12 12.20 60.00 58.39 1.92 6.40 65.00 66.09 1.17 8.00 65.00 64.03 1.06 10.80 65.00 64.36 0.70 75 °C 90 °C 105 °C 25.00 30.00 27.61 5.87 28.50 30.00 30.76 1.78 33.00 30.00 32.95 6.64 23.50 35.00 32.47 5.30 26.90 35.00 35.46 0.92 31.00 35.00 38.24 6.25 21.80 40.00 38.13 3.39 25.30 40.00 40.28 0.50 29.00 40.00 43.70 6.25 18.80 50.00 48.49 2.16 21.90 50.00 50.95 1.34 25.40 50.00 53.99 5.42 17.00 55.00 54.95 0.07 20.30 55.00 56.17 1.49 23.80 55.00 58.75 4.66 15.70 60.00 59.72 0.34 18.90 60.00 60.84 0.98 22.00 60.00 64.24 4.83 14.60 65.00 63.82 1.29 17.00 65.00 67.33 2.49 20.50 65.00 68.94 4.16

243 J.A. TAVARES et al.: CRYSTALLIZATION AND SEPARATION OF KCl… Chem. Ind. Chem. Eng. Q. 24 (3) 239−249 (2018)

Figure 2. Trajectory A: Part of the solubility diagram for the KCl-MgCl2-H2O system with saturated sodium chloride, recreated up to the equilibrium (curve E) of the solution with the carnallite and potassium chloride solids. Dissolution of carnallite at 25 °C.

Figure 3. Trajectory B: Part of the solubility diagram for the KCl-MgCl2-H2O system with saturated sodium chloride, recreated up to the equilibrium (curve E) of the solution with the carnallite and potassium chloride solids. Dissolution of carnallite at 35 °C. ture of 105 °C. In this stage, its concentration inc- chloride crystallizes, with recovery of approximately reases with water evaporation up to point 2 of the pro- 62% of the potassium chloride content. After removal cess, at which there is crystallization of around 47% of the crystals at point 3, the solution is subjected to a of the sodium chloride dissolve-d in the solution. Next, new evaporation of water in order to increase the con- the heated and concentrated solution is taken to a centration of the dissolved compounds. This evapor- crystallizer, where it is cooled to 25 °C, when it ation continues until reaching a temperature of 105 °C reaches point 3 of the system, at which the potassium at point 4 of the system, at which there is crystalliz-

244 J.A. TAVARES et al.: CRYSTALLIZATION AND SEPARATION OF KCl… Chem. Ind. Chem. Eng. Q. 24 (3) 239−249 (2018)

Figure 4. Trajectory C: Part of the solubility diagram for the KCl-MgCl2-H2O system with saturated sodium chloride, recreated up to the equilibrium (curve E) of the solution with the carnallite and potassium chloride solids. Dissolution of carnallite at 25 °C without crystallizing the synthetic carnallite in the process. ation of approximately 45% of the sodium chloride potassium contained in the synthetic carnallite is dissolved in the solution. After that, the heated 100%, representing an efficiency of approximately solution is concentrated and cooled until it reaches a 57% in relation to the feeding at point 5. In the crystal- temperature of 90 °C, at point 5 of the system, when lization of the potassium chloride, corresponding to crystallization of the potassium chloride occurs with points 2 and 5, a precipitation of 4%, at maximum, of an efficiency of around 20%, in relation to the solution the sodium chloride contained in the solution occurs, fed in the second evaporation at point 3 of the based on the information available in Ullmann’s system. Point 5 of the system represents the equi- Encyclopedia [3] and GEA [19]. At the end of the librium point between the solid potassium chloride three stages of crystallization of potassium chloride, a and the synthetic carnallite, according to the Hoffman total efficiency of slightly over 85% is obtained in chart [10]. From point 5 and after removal of the crys- relation to the feeding at the beginning of the system tallized potassium chloride, the solution is cooled until of crystallization, considering all the potassium chlo- reaching 25 °C, indicated by point 6, at which crystal- ride diluted in the salting solution at the beginning of lization of the synthetic carnallite occurs according to the system. the Hoffman diagram [10] and the van ’t Hoff theory The trajectory described above is one of the described in Ullmann’s Encyclopedia [3]. The crystal- possible alternatives for simulation of the process lization of synthetic carnallite represents a potassium using the equation developed from the Hoffman dia- chloride crystallization of approximately 57% in rel- gram [10]. It can also be seen in the flowchart in ation to the feeding at point 5 of the system. This syn- Figure 5. thetic carnallite crystallized between points 5 and 6 is Trajectory B (Figure 3) differs from trajectory A added to a solution with the same composition as the (Figure 2) only in the temperatures for dissolution of one used at the beginning of this system (point 1), carnallite (point 1) and synthetic carnallite (between which was saturated with potassium chloride. When points 7 and 8), and crystallization of the synthetic the synthetic carnallite is added, the magnesium chlo- carnallite (between points 5 and 6). Trajectories A ride is diluted, precipitating the potassium chloride and B are represented in the flowchart in Figure 5. until reaching point 8 (point 6) of the system, at which In trajectory C (Figure 4), the stages of crystal- dissolution of the synthetic carnallite cannot occur lization of synthetic carnallite (between points 5 and because the eutectic point of the solution is reached 6) are absent, while in the dissolution (between points at this temperature (25 °C). The efficiency of crystal- 7 and 8), as well as at point 3, there is a difference in lization of the potassium chloride in relation to the

245 J.A. TAVARES et al.: CRYSTALLIZATION AND SEPARATION OF KCl… Chem. Ind. Chem. Eng. Q. 24 (3) 239−249 (2018)

Figure 5. Flowchart of the crystallization process of potassium chloride in the first and second cores. a: Carnallite well; b: Salting tank; c: Evaporator; d: Crystallizer; e: Centrifuge; f: Dryer; g: Heat exchanger and Furnace; h: Evaporator; i: Crystallizer; j: Centrifuge; k: Crystallizer; l: Centrifuge; m: Crystallizer; n: Centrifuge; o: Dryer; p: Dryer. the cooling temperature, with cooling proceeding until Economic feasibility of the process reaching 55 °C. The procedure for evaluation of the economic Process flowchart feasibility of constructing and operating a plant for crystallization of potassium chloride was designed in The flowchart was created and developed based accordance with all the technical parameters avail- on the processes available in the literature and the able in the literature, employing the equations deve- solubility properties of the components, mainly car- loped to accelerate the potassium chloride crystalliz- nallite, can be divided into two cores. ation process, relative to the Hoffman diagram [10]. The first core consists of the drilling well, to Eq. (6) was used to calculate the equipment costs, which water is added in order to dilute the carnallite considering the initial cost and capacity of each item and generate the salting solution. The solution is then of equipment. The exponent of this equation refers to pumped out and stored in tanks, in order to be contin- the type of equipment for which the cost will be cal- uously fed into an evaporator where the salts are culated. In the absence of this exponent for any par- concentrated by evaporation. After the evaporator, ticular item, the rule of the six-tenths factor was used, the concentrated salting solution undergoes crystal- which can provide satisfactory results when applied to lization by cooling, with precipitation of potassium any equipment [11,13–16]: chloride and its subsequent separation from the salting solution in a centrifuge. The salting solution that Cost of equip. a = leaves the centrifuge is taken to the second core of n Capac. equip. a the process. The potassium chloride separated in the = Cost of equip. b  Capac. equip. b centrifuge is sent to the drying process, in which  water is removed from the salt. Together with the pre- Hence, it was possible to assess the economic cipitation of potassium chloride, a small amount of feasibility of the process according to the capacity sodium chloride is precipitated, which is diluted in the used in the simulation of the process, with automatic salting solution. A detailed flowchart of the processes calculation of the cost of each item of equipment and in the first and second cores is presented in Figure 5. all other costs described in the methodology. Using This flowchart is based on trajectories A and B (Fig- these procedures, many values for the production of ures 2 and 3). potassium chloride were determined hourly and yearly, together with the corresponding annual rates of return and the overall times for investment return,

246 J.A. TAVARES et al.: CRYSTALLIZATION AND SEPARATION OF KCl… Chem. Ind. Chem. Eng. Q. 24 (3) 239−249 (2018) considering the amount of carnallite fed in the pro- values for the processes developed in this work. The cess. Table 3 shows the values for the traditional pro- chart of Figure 6 compares the rates of return and the cess (without the second core part), which has been annual productions of potassium chloride for traject- used for many decades, while Tables 4-6 show the ories A, B and C.

Table 3. Economic analysis for different feedings of potassium chloride, using the traditional process of crystallization by cooling of the concentrated salting solution

Crystallization Feeding, t of KCl/h Return rate, %/year Investment return, year Total efficiency, % t of KCl/h 103 t of KCl/year 0.50 65.41 0.31 2.45 -12.74 - 1.00 0.62 4.90 -6.92 - 2.00 1.24 9.79 0.48 210.51 5.00 3.09 24.5 12.62 7.92 10.0 6.18 49.0 23.35 4.28 20.0 12.4 97.9 35.01 2.86 50.0 30.9 245 50.98 1.96 100 61.8 490 62.75 1.59 200 124 979 73.69 1.36 400 247 1,959 83.42 1.20 500 309 2,448 86.24 1.16 1,000 618 4,897 93.96 1.06

Table 4. Economic feasibility for different feedings of potassium chloride, using the process developed in this work for trajectory A, shown in Figure 2

Crystallization Feeding, t of KCl/h Return rate, %/year Investment return, year Total efficiency, % t of KCl/h 103 t of KCl/year 0.50 85.18 0.43 3,37 -10.98 - 1.00 0.85 6.75 -4.98 - 2.00 1.70 13.5 2.46 40.59 5.00 4.26 33.7 14.48 6.91 10.0 8.52 67.5 24.99 4.00 20.0 17.0 134 36.35 2.75 50.0 42.6 337 51.89 1.93 100 85.2 675 63.35 1.58 200 170 1,349 74.02 1.35 400 341 2,698 83.50 1.20 500 426 3,373 86.25 1.16 1,000 852 6,746 93.71 1.07

Table 5. Economic feasibility for different feedings of potassium chloride, using the process developed in this work for trajectory B, shown in Figure 3

Crystallization Feeding, t of KCl/h Return rate, %/year Investment return, year Total efficiency, % t of KCl/h 103 t of KCl/year 0.50 83.39 0.40 3.18 -11.69 - 1.00 0.80 6.36 -5.87 - 2.00 1.61 12.7 1.43 69.86 5.00 4.02 31.8 13.39 7.47 10.0 8.04 63.7 24.01 4.16 20.0 16.1 127 35.69 2.80 50.0 40.2 318 51.97 1.92 100 80.4 637 64.23 1.56

247 J.A. TAVARES et al.: CRYSTALLIZATION AND SEPARATION OF KCl… Chem. Ind. Chem. Eng. Q. 24 (3) 239−249 (2018)

Table 5. Continued

Crystallization Feeding, t of KCl/h Return rate, %/year Investment return, year Total efficiency, % t of KCl/h 103 t of KCl/year 200 83.39 161 1,273 75.84 1.32 400 322 2,546 86.35 1.16 500 402 3,183 89.43 1.12 1,000 804 6,366 97.93 1.02

Table 6. Economic feasibility for different feedings of potassium chloride, using the process developed in this work for trajectory C, shown in Figure 4

Crystallization Feeding, t of KCl/h Return rate, %/year Investment return, year Total efficiency, % t of KCl/h 103 t of KCl/year 0.50 83.27 0.39 3.12 -13.28 - 1.00 0.79 6.23 -8.22 - 2.00 1.57 12.5 -1.86 - 5.00 3.94 31.2 8.59 11.65 10.0 7.87 62.3 17.94 5.57 20.0 15.7 125 28.29 3.54 50.0 39.4 312 42.86 2.33 100 78.7 623 53.94 1.85 200 157 1,247 64.52 1.55 400 315 2,493 74.17 1.35 500 394 3,117 77.01 1.30 1,000 787 6,233 84.88 1.18

Figure 6. Comparative chart of return rates versus the production of potassium chloride for trajectories A, B, and C.

CONCLUSION ions by cooling were performed: in the first, the potassium chloride crystallized with an efficiency of 6.5% in The final process proposed presented excellent relation to the potassium chloride fed in the first core, results, with over 85% of the potassium chloride crys- while in the second there was crystallization of the tallized (trajectory A). In the first core of the process, synthetic carnallite. To this crystallization was added approximately 62% of the fed potassium chloride a third crystallization in the second core, using a salt- crystallized. In the second core, two other crystallizat- ing solution under the same feeding conditions as the

248 J.A. TAVARES et al.: CRYSTALLIZATION AND SEPARATION OF KCl… Chem. Ind. Chem. Eng. Q. 24 (3) 239−249 (2018) first core, with approximately 17% of all the potassium de Produção Mineral, Brasília, 2014, pp. 100-101 (in Por- chloride fed in the process. tuguese) nd The process simulation of crystallization of pot- [5] J.K. Warren, Evaporites A Geological Compendium. 2 assium chloride showed that trajectory A (Figure 2) ed., Springer, Berlin, 2016 achieved better results, compared to the other traject- [6] S. Abraham, Israel Chemicals Ltd., US4140747 (1977) ories (Figures 3 and 4), as can be seen from the [7] A.F. Nylander, Kaiser Aluminum Chem. Corp., comparison of the results shown in Tables 5-7, and US3642454 (1972) Figure 6. A greater production of potassium chloride [8] N. Gruschow, F.K.H. Walkhoff, W. Ulrich. Deusa Projekt was obtained, with consequently greater profitability, Management e Kavernen Bau-Und Betriebs, US6022080 (1997) even having a smaller annual return rate for large- [9] R.Phinney, Karnalyte Resources Inc., US8282898 (2009) scale potassium chloride processing. There was also [10] H.G. Hoffman, Kristallisation in der industriellen, Praxis, a smaller amount of waste generated, relative to the Wiley-VCH, Weinheim, 2004 crystallized potassium chloride. [11] R.K. Sinnott, Chemical Engineering Design, 4th ed., Acknowledgments Elsevier Butterworth-Hienemann, New York, 2005 [12] D.M. Himmelblau, J.B. Riggs, Basic principles and calcul- Financial support for this work was provided by ations in chemical engineering, 7th ed., Prentice Hall, the Brazilian agency Coordination for the Improve- Upper Saddle River, NJ, 2004 ment of Higher Education Personnel (CAPES). The [13] Matches, Matches' Process Equipment Cost Estimates, authors would like to thank the former coordinator of http://www.matche.com/equipcost/Default.html (accessed the Graduate Program in Chemical Engineering of 22 Jun 2015) UFS Car, Dr. José Maria Corrêa Bueno, for his enc- [14] M.S. Peters, K.D. Timmerhaus, Cost Estimator Tool, ouragement in undertaking this research. http://www.mhhe.com/engcs/chemical/peters/data/ce.htm l (accessed 22 Jun 2015) REFERENCES [15] W.D. Baasel, Preliminary Chemical Engineering Plant Design, 2nd ed., Van Nostrand, New York, 1990 [1] M. Nascimento, M.B.M. Monte, F.E. Lapido-Loureiro. [16] M.S. Peters, K.D. Timmerhaus. Plant Design and Eco- Agrominerais - Potássio. In: A.B. Luz, F.A.F. Linz, nomics for Chemical Engineers, 5th ed., McGraw-Hill, nd Rochas e Minerais Industriais: Usos e Especificações, 2 New York, 2003 ed., CETEM, Rio de Janeiro, 2008, pp. 175-209 (in [17] The Economist, Big Mac Index, Portuguese) http://www.economist.com/content/big-mac-index [2] P.W. Harben, The Industrial Mineral Handybook – A (accessed 22 Jun 2015) Guide to Markets, Specifications and Prices, 4th ed., [18] Chemical Engineering, Chemical Engineering Plant Cost Industrial Minerals Information Services, London, 2002 Index (CEPCI), http://www.chemengonline.com/issues/ [3] H. Shultz, G. Bauer, E. Schachl, F. Hagedon, P. Schmit- /2015-06 (accessed 22 Jun 2015) tinger, Potassium Compounds, Ulmann’s Encyclopedia of [19] GEA Messo PT, Potash Fertilizer Industry, http:// Industrial Chemistry, V. A22, VCH-Verlag GmbH & Co, //www.gea.com/en/binaries/2010-02_Potash%20fertilizer- Weinheim, 1993 %20industry_tcm11-21927.pdf (accessed 12 Feb 2013). [4] L.A.M. Oliveira, Potássio, in: Sumário Mineral 2014, T.M. Lima, C.A.R. Neves, Eds., Brasil Departamento Nacional

J.A. TAVARES KRISTALIZACIJA I SEPARACIJA KCl IZ RUDE L.F. MOURA KARNALIT: RAZVOJ PROCESA, SIMULACIJA I A. BERNARDO EKONOMSKA OPRAVDANOST M. GIULIETTI

Chemical Engineering Department, S obzirom na sve veću potražnju za kalijumom u Brazilu, kopanje i upotreba karnalita Federal University of São Carlos postaju sve važniji, jer se trenutni izvor kalijuma (silvinit) iscrpljuje pa postoji rizik od (UFSCar), Rodovia Washington Luiz, nestašice. Na osnovu teorijskih i praktičnih podataka dostupnih u literaturi, ovaj rad opi- São Carlos, SP, Brazil suje razvoj, simulaciju i ekonomsku izvodljivost postupka za rastvaranje i kristalizaciju kalijum hlorida iz karnalitne rude. Pozitivni rezultati dobijeni su nakon primene Hoffman NAUČNI RAD dijagrama i određivanja odgovarajuće jednačine. Predloženi postupak je obezbedio kristalizaciju preko 85% kalijum-hlorida, pokazujući svoje superiorne performanse, u pore- đenju sa postojećim procedurama.

Ključne reči: karnalit, kalijum-hlorid, kristalizacija, proces sinteze, ekonomska procena.

249