A Novel Procedure for Performance Assessment of Paste Thickeners

ABSTRACT High stability and low settling rate of phosphate fines in water suspensions encountered the dry area-located mines with serious environmental and technical challenges. The best suggestion for these mines is paste thickener installation for dewatering. In this research, the novel procedure has been proposed for evaluating the potential of paste thickeners in dewatering mining tailings. The case study is Esfordi phosphate complex (Iran). Initially, the selection criteria of a proper flocculant (environmental effects, price, particle settling rate, turbidity of overflow water, and underflow water content) were evaluated for current consuming flocculant (A-27) and 12 other ionic and nonionic flocculants. Then, flocculation performance was ranked based on the type and concentration, applying multiple criteria decision-making (MCDM) techniques of ORESTE, MAPPAC, and ELECTRE. Subsequently, the pilot-scale tests were performed for the selected flocculants and the operational conditions were evaluated for the maximum underflow solid content of 70 wt.%. The results of the present research showed that installation and application of this type of thickener could be effective for the case study.


Introduction
The problems associated with phosphatic clay management have encouraged the researchers to find economical and practical methods for dewatering and consolidating.These methods were documented in the literature (Boshoff, Morkel and Naude 2018;El-Shall and Zhang 2004;Garmsiri and Nosrati 2019;Gnandi et al. 2005;Kong and Orazem 2014;Ozacar and Ayhan Sengil 2003;Scheiner and Smelley 1985).
Although, storing phosphate tailings in clay ponds has been the most common method for disposal but now a more effective approach is followed by the industries to cover issues including a huge amount of water loss, occupying large pieces of land, and the possibility of environmental disasters (Besra, Sengupta and Roy 1998;Rulyov, Dontsova and Korolyov 2005;Tao, Parekh and Honaker 2008).
According to a research carried out for dewatering disposal in Esfordi phosphate complex (Iran), the lowest cost method for tailings disposal considering environmental conditions of the area is reported to be tailings dewatering by paste thickeners and transferring the slurry by capable pumps to the dam.Consequently, replacing conventional thickeners with paste thickeners (also known as deep cone thickeners) was chosen as an effective practical approach.Considering the importance of the subject, this research was conducted in two parts to investigate the feasibility of using paste thickener for the complex.
The first step to achieve optimum dewatering results is choosing the type and concentration of the flocculant in batch scale experiments.These flocking agents depending on the type and consuming dosage significantly influence particle settling rate, turbidity of overflow water, and underflow water content.Although, they can impose both operational cost and environmental effects on the mineral processing industry.In this research, the mentioned criteria were evaluated for the current flocculant of the complex (A-27) and 12 other ionic and nonionic flocculants.Obviously, an effective flocculant was identified with low consuming dosage, high particle settling rate, low turbidity of overflow water, and low underflow water content.Therefore, given the multiplicity of conflicting criteria and necessity of choosing the effective type and concentration of flocculants, multiple criteria decision-making (MCDM) techniques were applied in this study.MCDM techniques used in the mining research were mainly focused on its environmental effects and less on mineral processing.A summary of this research is given in Table 1.
In this research, to determine the type and concentration of the flocculant with respect to five criteria, MCDM techniques of ORESTE, MAPPAC, and ELECTRE (supplementary file) were applied for the first time in mineral processing field.The validation of ranking results was carried out using the three techniques.Precise selection of type and concentration of flocculant improves the results of subsequent pilot settling experiments and enhances the capability of paste thickener for this complex.The novel procedure proposed in this research can be considered as a basis for evaluating the potential of paste thickeners in dewatering of mining tailings.

Ore sample
The ore sample was received from Esfordi phosphate complex taken from the entrance launder of pulp to the thickener.The ore was filtered and dried.Characterization of the sample by X-ray powder diffraction (XRD) revealed that hematite, quartz, and fluorapatite were major phases, while, chlorite, talc, hornblendes, calcite, orthoclase, and dolomite were minor phases.Particle size distribution was measured by a particle size analyzer (Fritsch, "ANALYSETT 22").Accordingly, arithmetic and geometric mean diameters of the particles were calculated as 10.7 μm and 6.5 μm, respectively, indicating the obstacle in settling.The natural settling rate of ore was obtained as 0.0066 cm/min.

Flocculants
Flocculation occurs by polymer bridging, charge compensation or neutralization, polymer surface complex formation and depletion, and/or by a combination of these mechanisms (Mpofu, Addai-Mensah and Ralston 2004;Mpofu, Mensah and Ralston 2003;Onen, Beyazyuz and Yel 2017).This process causes an increase in the settling rate, the improvement of overflow clarity and the increase in the thickener capacity by decreasing particle residence time.Polymers are widely used flocculants that can be synthesized in various molecular masses and ionic forms (Gregory and Barany 2011;Kim and Palamino 2009).Consumed flocculants in this research were cationic (DHMW, EHMW, GHMW, and BHMW), anionic (A-26, A-27, A-100, A-110, A-130, A-150, X0, and X1), and nonionic (N-100).A-27 (Isfahan Copolymer Co., Iran) is the current consuming flocculant in the Esfordi phosphate complex.

Settling experiments
Settling experiments were conducted in a graduated cylinder with a capacity of 1000 ml and 35 cm of height.A 5% w/v slurry was provided for each settling experiment similar to the current solid percentage of the industrial thickener set.The normal pH of the slurry was measured as 7.9.Then, flocculants with the concentrations of 2, 4, 5, and 7 g/t were added and were mixed well by moving a mixer up and down equipped with a net-like circular plate at a slow rate.To calculate the settling rate, bed height changes were recorded as a function of time.After conducting settling experiment, overflow water was completely rejected and underflow was weighed, and then was subjected to a vacuum filter for a specified time.Water content was calculated based on the ratio of water weight of underflow to the whole weight of  Settling experiments were carried out using a pilot-scale paste thickener, schematic diagram and dimensions of which is shown in Figure 1a.This thickener was constructed by colorless Plexiglas to monitor the exact settling behavior, characterizing different zones including clear water, hindered settling, and compression and also determining pulp solid content profile in the zones.Furthermore, it contributes in evaluating the paste flow behavior from underflow and the effects of operational variables such as bed height and residence time on underflow solid content.Some sampling faucets with determined distances were embedded on the thickener wall.This thickener is fed with a height adjustable feed well designated for mixing the pulp with flocculant prior to introducing the settling part (Figure 1b).The height adjustability of the feed well avoids short circuiting and interference of overflow clear water with introducing pulp.
Pilot settling experiments were performed in both continuous and non-continuous state.In the non-continuous state, pulp and flocculant solution were introduced to the thickener at 3 m height through the feed well, where the thickener was filled with water.Then, bed line height was measured at different times, until it reached up to 3 m height from the bottom of settling column.In the continuous state, the thickener was fed using the noncontinuous state to reach the desired level.Then, feeding process continued for 5 h to achieve steady state, in other words, to have the same rate of solid content in both feeding and underflow.

Slump test
The slump test is used extensively by the engineers to estimate the yield stress and viscosity of paste (Tao, Parekh and Honaker 2008).This test was carried out using a PVC cylinder with height to diameter ratio of 1:1 under different conditions.First, paste with solid contents of 65%, 70%, and 75% was prepared using a mixture containing 5 g/t A-130 flocculant, solid sample and water.The cylinder was placed on a smooth surface and prepared pastes with different solid contents were added until it was filled.Then, the cylinder was taken upward slowly and the paste was let to flow.
As shown in Figure 2, the difference between the initial and final heights (center point of collapsed paste) was measured as a slump.By measuring the density of produced pastes, yield stress (τ) could be calculated as in Eq. ( 1) (Tao, Parekh and Honaker 2008): where S is the difference between initial and final height (m), H is the initial height (m), g is gravitational acceleration (m/s 2 ), and ρ is paste density (kg/m 3 ).

Selection of flocculant type and concentration
In this research, the current flocculant used in the plant (A-27) and 12 other ionic and nonionic flocculants were evaluated with respect to the above criteria.Then, flocculants were ranked according to the type and concentration using MCDM techniques of ORESTE, MAPPAC, and ELECTRE.
Regarding the calculations in MCDM techniques, the formation of decision matrix (R) is the first step.In this matrix, the element assigned to alternative i and criterion j is represented by r ij so, this element indicates the characteristic of alternative i in the viewpoint of criteria j. 13 flocculants in 4 concentration levels were considered as the alternatives in the study defined as A1 to A52.Environmental effects of flocculants based on the consumption dosage (g/t), flocculant price (cent), particle settling rate (cm/min), turbidity of overflow water (NTU), and underflow water content (%) were evaluated as criteria defined as X1 to X5, respectively.Among all criteria, solely a decrease in the settling rate had a positive effect.The concentrations of flocculants were chosen as the value representing environmental effect for each alternative.Decision matrix was formed (Eq.(S-29)) with respect to alternative values for each criterion, as presented in the supplementary file.Due to the large volume of calculations and the large number of tables, only a few of them are presented in the following.
In the following, relative importance of criteria was determined according to experts' experience.For this purpose, 10 experts' opinions were collected using the ranking method where terms such as the least importance, less importance, mean importance, high importance, and the highest importance were assigned to 1, 2, 3, 4, and 5, respectively.The average of experts' opinions was determined for the final weight, and then was normalized.In fact, the final weight is the ratio of the average of experts' opinion for each criterion to sum of experts' opinion averages for five criteria.Accordingly, the weights of 0.12, 0.26, 0.37, 0.09, and 0.16 were assigned to criteria of X1, X2, X3, X4, and X5, respectively.

Calculation stages in ORESTE technique
3.1.1.1.Primary ranking of alternatives.First, two preference structures were proposed for a set of criteria and alternatives in the decision matrix.The weights determined by the experts were applied to create the preference structure for criteria, and the second preference structure on the alternatives was similarly set based on each criterion.Then, Besson's average ranking method was applied to the initial ranking of criteria and alternatives set (Eq. ( 2)).3.1.1.3.Global ranking of distances.In this step, first, the prior results were ranked by Besson's average ranking method to achieve global ranks (R(m k )).Then, R(m) with value equal to the sum of R(m k ) was calculated for each alternative (Eq.( 4)).Finally, ranking of alternatives was performed according to R(m) results as shown in Table 2.In ORESTE technique, a better ranking is given to the alternative with smaller R(m).

Calculation steps in MAPPAC technique
3.1.2.1.Determination of normalized decision matrix, C. In the first step of ranking, using these technique ideal values of 5, 2.86, 13.34, 321.9, and 5.76 and base values of 0, 0, 2.99, 0, and 0 were given to criteria of X1 to X5, respectively.After determination of ideal and base values, normalized decision matrix was calculated and presented as shown in Eq. ( 5).
3.1.3.5.Formation of effective general matrix.The effective whole (general) matrix is given in Eq. ( 13), which is the product of concordance matrix multiplied by discordance matrix.
3.1.3.6.Alternatives ranking in ELECTRE technique.In ELECTRE technique, an alternative is favorable and preference when the column and row of effective general matrix has a maximum value of 0 or 1, respectively.In this study, each alternative score was determined by the summation of elements of each row, and preference alternatives were chosen according to the obtained scores.Table 4 shows the ranking of alternatives.

Final alternatives ranking
All three techniques of ORESTE, MAPPAC, and ELECTRE were used in alternatives ranking to validate the results.The comparative evaluation of the results showed that alternatives 28 and 35 allocated the best ranks among the others using all techniques, respectively.The best results in settling experiments were concluded for studied criteria using 5 g/t of A-130 and N-100 flocculants.Regarding the results of ranking difference for other alternatives and also for common ranking between them, the arithmetic average was carried out for ranking results of alternatives ranking (Table 5).As shown in Table 5, the current consuming flocculant in the complex (A-27) at concentrations of 2, 4, 5, and 7 g/t placed in ranks of 9, 17, 39, and 42, respectively.

Evaluation of pilot settling experiments for non-continuous feeding
Settling flux is calculated using Eq. ( 14), which is a common criterion to select optimum feed solid content in the thickener.In thickener designing, the feed solid content with the highest settling flux is selected for implementation.The settling flux is calculated from Eq. ( 14): where Gs is settling flux (t/h/m 2 ), U(c) is pulp settling rate (m/h), and C is pulp concentration (t/m 3 ).
Pilot settling experiments were done using N-100, A-130, and A-27 at a dosage of 5 g/t and settling flux of 2.0, 1.8, and 1.6 (t/h/m 2 ). Figure 3 shows the changes in bed height as a function of time.As shown in Figure 3, while using A-130, N-100, and A-27 flocculants, 3-m height bed was set in 32, 40, and 46 min, respectively.These results are in good agreement with bench scale settling experiments.Based on Eq. (S-29) presented in supplementary file, settling rates obtained in regard to their flocculant agents are in the following order: A-130 > N-100 > A-27.Due to higher settling rate of 5 g/t A-130 flocculant, pilot settling experiment was performed at settling flux of 2 t/h/m 2 .In this experiment, a 5 wt.% pulp was introduced at 3 m of height to the thickener and by adjusting bed height at 1 and 2 m, sampling from underflow was performed within 1, 2, 3, 5, 7, and 9 h. Figure 4 shows the effect of bed height on underflow solid content.
The flocs generated from mixing the pulp with flocculant solution in hindered settling zone were gradually contracted through hydrostatic pressure of higher zones as well as the pressure of settling particles layers.As illustrated in Figure 4, the contraction is maximized after 5 h and the most changes of underflow solid content occur in the time range of 1 to 3 h.Besides, by increasing bed height from 1 to 2 m, pressure on layers increases in hindered settling zone.Therefore, underflow solid content clearly increases as a result of water releasing from flocs.In this research, the maximum underflow solid content was achieved at bed height of 1 and 2 m by 58% and 70%, respectively, denoting successful performance of pilot-scale paste thickener in dewatering of Esfordi phosphate tailing.

Evaluation of pilot settling experiments in continuous feeding
Under optimum conditions obtained from pilot settling experiments in non-continuous feeding (flocculant of A-130, at a dosage of 5 g/t, bed height of 2 m, time of 5 h, and settling flux of 2 (t/h/m2)), pilot settling experiment was carried out in continuous feeding at 3 m of feeding height for 5 h.Sampling was performed at 1 h intervals from all faucets.The solid percentage in different times and bed heights is given in Table 6.
The results of average solid percentages at different heights (Table 6) indicated that the increase in the solid percentage occurs at 1 to 2 m of height from underflow discharge as a result of the compression process.Within 1 to 2 m of distance from underflow discharge, the increase in the solid percentage rate changes with slower pace due to the declined water content in flocs.
Totally, the basis of particle settling in thickeners is the simple model represented in Figure 5, in which it is divided into three distinct zones including clarification, hindered settling, and compression zones vertically.
The three determined zones can be defined for pilot thickener based on the profile given in Figure 5.The clarification zone is located at 2.25 to 4.0 m of height.The height range between 1.0 and 2.25 m from underflow discharge is allocated to hinder settling.Figure 6 shows the changes in the solid percentage in this zone.Compression zone (paste formation) is located at 0.0 to 1.0 m of height from the bottom of settling column.The compression zone could be separated into two parts.In the first part, pulp turns into a high concentration slurry (0.5 to 1.0 m).The second part of compression zone is related to the height ranging from 0.0 to 0.5 m from discharge.Figure 7 shows the changes of the solid percentage in this zone.The compression zone is the fundamental difference between conventional and paste thickeners evacuating the maximum water to change the high concentration pulp into a paste.

Evaluation of yield stress
The paste has a solid-like behavior as concentration is one of its characteristics.Unlike the pulp which principally is liquid and is formed in the shape of its container, paste shape is usually constructed based on its concentration.Pressure is required to flow the paste.This pressure is a rheological terminology recognized as yield stress, which is a unique feature of non-Newtonian fluids.
Within compression process in industrial thickeners, by increasing solid concentration the yield stress increases, consequently paste flow turns into a difficult task.Therefore, regarding the paste thickeners, the increase in the underflow solid content is not the sole criterion for choosing variables conditions.As discussed earlier, the increase in the solid percentage and yield stress results in a need for higher driving force for rakes leading to the disruption in pump performance to transport paste to the tailings dam.
The slump test is carried out to estimate the yield stress, flow, and competency of paste for disposal.In this research, after measuring the paste slump produced by 5 g/t of A-130 flocculant, yield stress was calculated by Eq. (1). Figure 8 shows the changes in yield stress as a function of paste solid percentage.As shown in Figure 8, yield stress increased significantly.For yield stress, the change is calculated as 250 Pa with the increase in the solid percentage from 65% to 75%.
The results of a study showed that (Wills 2011), the yield stress of 175 to 300 Pa along with a solid percentage of 70-75% is suitable for paste flow and transportation purposes.Hence, design, construction, and implementation of a paste thickener were feasible in the Esfordi phosphate complex.

Conclusion
Regarding the feasibility of applying paste thickeners for mineral processing units, in this research a novel procedure was presented for assessment of this equipment in dewatering of phosphate tailings.Firstly, three various MCDM techniques including ORESTE, MAPPAC, and ELECTRE were applied to choose type and concentration of proper flocculants for a paste thickener according to environmental effects of flocculants, flocculant price, particle settling rate, turbidity of overflow water, and underflow water content.The results showed that the first (A-130, 5 g/t) and second (N-100, 5 g/t) ranks are similar for all three techniques indicating validated results.It is highly recommended to use A-130 and N-100 instead of A-27 for further experiments in designing the paste thickener.The results of experiments for measuring bed height in non-continuous feeding state revealed that A-130, N-100, and A-27 flocculants require fewer times to form bed height at 3 m of height from underflow discharge.In case of A-130 flocculant (5 g/t), the results indicated that, 70% of underflow solid percentage was obtained at 2 m of height, at 5 h of residence time, and settling flux of 2 (t/h/m 2 ).Under these conditions for continuous feeding state, clarification,  hindered settling, and compression zones were determined at 2.25-4.0,1.0-2.25,and 0.0-1.0m, respectively.Finally, according to the slump test results, the yield stress between 175 and 300 Pa indicates that the paste can flow and be transported by the pumps.The results of the research are promising for solving some environmental effects and lack of water for the Esfordi phosphate plant in case of employing a paste thickener.
and its dose rate in the flotation of lead-zinc sulfide ore TOPSIS Kostovic and Gligoric (2015) Selection of alunite processing method based on 14 technical, economic and environmental analyzing criteria DAHP & Fuzzy TOPSIS Alizadeh, Salari Rad and Bazzazi (2016) underflow.A WTW 655 IR turbidity meter was exploited to measure the turbidity of overflow water.The results of batch scale settling experiments for settling rate, turbidity, and water content are presented in Figures S-1, S-2, and S-3, respectively, in the supplementary file.

Figure 1 .
Figure 1.Schematic diagram and dimensions of the designed paste thickener (a) and feed well (b).

Figure 2 .
Figure 2. The measurement method of slump.

Figure 3 .Figure 4 .
Figure3.Effect of the chosen flocculants on bed height in non-continuous feeding.

Figure 6 .
Figure 6.Changes of the solid percentage in the zone of hindered settling (1.0 to 2.25 m).

Figure 7 .Figure 8 .
Figure 7. Changes of the solid percentage in the zone of compression settling (0.0 to 1.0 m).

Table 1 .
Summary of the environment and mineral processing research using MCDM techniques.
Calculation of alternatives distances.In this study, direct linear estimation method was used to estimate distances in which the corresponding matrix is presented as shown in Eq. (3).

Table 2 .
Alternatives ranking by ORESTE technique.

Table 5 .
Final ranking of alternatives based on averaging from ranking results of all three techniques.

Table 6 .
The solid percentage of samples in continuous feeding.