Wall effects during settling in cylinders

This article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal non-commercial research and education use, including for instruction at the authors institution and sharing with colleagues. Other uses, including reproduction and distribution, or selling or licensing copies, or posting to personal, institutional or third party websites are prohibited. In most cases authors are permitted to post their version of the article (e.g. in Word or Tex form) to their personal website or institutional repository. Authors requiring further information regarding Elsevier’s archiving and manuscript policies are encouraged to visit: http://www.elsevier.com/authorsrights Author's personal copy Colloids and Surfaces A: Physicochem. Eng. Aspects 449 (2014) 157–169 Contents lists available at ScienceDirect Colloids and Surfaces A: Physicochemical and Engineering Aspects journal homepage: www.elsevier.com/locate/colsurfa Wall effects during settling in cylinders Benjamin Buratto, Shane P. Usher ∗ , David Parris, Peter J. Scales Particulate Fluids Processing Centre, Department of Chemical and Biomolecular Engineering, The University of Melbourne, Victoria 3010, Australia h i g h l i g h t s • Compared settling of alumina slurries g r a p h i c a l Rod Cylinder in cylinders and cylinders with vertical rods. • Polymer flocculated solids shrank from rod and settled faster: Shear densification. • Salt coagulated solids clung to the rod and settled slower: Syneresis. • Aggregation state influences settling rate changes in presence of vertical rods. Suspension a r t i c l e a b s t r a c t i n f o Article history: Received 4 September 2013 Received in revised form 16 February 2014 Accepted 17 February 2014 Available online 28 February 2014 Keywords: Densification flocculation settling tests syneresis wall effects Liquid A - Syneresis B – Shear densificaon The phenomena of syneresis and shear densification were analysed through the sedimentation of alumina suspensions. Experiments compared cylinder settling tests with a rod inserted in a cylinder against free settling without a rod for the cases of polymer flocculated and salt coagulated suspensions. Additionally, dye intrusion was used to visualise the final bed structure to identify the influence of cylinder walls and the potential presence of aggregate arching structures. It was found that, for various cylinder diameters, polymer flocculated alumina suspensions settled more quickly when a rod was placed in the centre of the settling cylinder. This is attributed to shear induced densification of the flocculated aggregates during the settling process, as it was observed that the solids shrank away from the rod, creating a channel which allowed water to escape more easily than when the rod was absent. The reverse result was observed when a rod was placed in an unflocculated/coagulated alumina suspension; the solids stuck to the rod, resembling the syneresis mechanism of isotropic shrinkage towards the rod. Industrially these results are significant since inserting rods into sedimentation devices such as a thickener can result in a flocculated suspension settling further and at a faster rate, such that rods can provide an alternative to mechanical devices such as rakes. Dye intrusion tests indicated that there was arching in the network bed, resulting in small pockets of water being trapped in the structure. An equation was derived to predict the maximum diameter arch that could be supported, taking into account the shear yield stress of the suspension. The solution to this equation was supported by observed cavities at the cylinder wall of up to 3 mm. These results suggest that cylinder walls may have a negative effect on the extent to which a suspension dewaters. © 2014 Elsevier B.V. All rights reserved. 1. Introduction Solid-liquid separation techniques are used extensively in industry in a variety of applications. There are two main ∗ Corresponding author: Tel.: +61 3 8344 5592; fax: +61 3 8344 4153. E-mail address: [email protected] (S.P. Usher). http://dx.doi.org/10.1016/j.colsurfa.2014.02.045 0927-7757/© 2014 Elsevier B.V. All rights reserved. a b s t r a c t mechanisms through which separation occurs; filtration and sedimentation. Filtration is based on separation by a physical barrier, as the liquid moves through the filter the solids are impeded. Sedimentation is the movement of solids due to a density difference under the influence of a body force such as gravity or centrifugal acceleration. The focus herein is on sedimentation as it occurs in thickeners and clarifiers typical of the minerals industry with the addition of flocculants to aggregate fine particles and achieve Author's personal copy 158 B. Buratto et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 449 (2014) 157–169 significant increases in solids volume fraction. Through the process of thickening, slurry is separated into an underflow of a higher solids concentration along with an overflow liquor (supernatant) that is effectively free from solids. Ideally, optimised gravity thickener operation involves stable performance that consistently achieves high underflow solids concentrations with minimal solids in the liquor overflow. In order to improve understanding of the continuous sedimentation process used in industry, batch bench-top sedimentation experiments are often performed in the laboratory [1]. A suspension of known solids concentration is allowed to settle under gravity in a cylinder and the height of the suspension-supernatant interface is recorded as it varies with time. Plotting and analysing the height versus time data can reveal information about the settling characteristics of the slurry [2,3]. These tests have been used widely to study aspects such as the effects of different flocculants [4–6] and the improvement of dewatering using rakes [7]. One aspect of dewatering performance which has not been extensively considered is the influence of surfaces and walls. Surfaces in the form of walls, sloped bases, lamellas, pickets, rods and rakes play important roles in improving dewatering performance [8]. Wall effects, including wall friction and adhesion to the wall, might be expected to make testing in relatively small diameter cylinders an unreliable guide to the behaviour of large bodies of sediment, but careful work using cylinders of different diameters can identify when wall effects are negligible for practical purposes. In this study we intend to focus on the settling region next to the wall and attempt to understand the underlying mechanisms at work. It is well known that once beds settle beyond their gel point, a gap forms between the solids and the wall at the top of the bed. This has been shown to be due at least in part to the wall supporting the bed adjacent to the wall; the centre of the bed then settles faster and draws the edges inwards, giving the characteristic ring shape on the top of the bed. What is not clear from this is the relative contribution of gravity and syneresis. Syneresis does not appear to have a single agreed definition. It actually defines a phenomenon rather than a mechanism, namely the shrinkage of gels over time with the expulsion of water. For the purposes of this paper it is defined as the shrinkage of a reticulated gel over time driven by the reduction in internal solid surface area and surface energy. Shrinkage due to applied pressure, such as the weight of the overhead portion of the bed, is excluded. Vliet et al. (1991) [9] refers to the former as endogenous syneresis. A review of a spectrum of work carried out on syneresis is included below. In a fully reticulated, uniform bed, such as one formed by fine coagulated alumina, the effects of syneresis can be predicted with some confidence. The bed will shrink isotropically, and the rate of shrinkage will depend on the size and the shape of the bed. The first effect is self-evident. The size and shape have an impact because the water has to percolate through the bed to be expelled, and the distance that water has to travel and the flux both depend on the size and shape of the bed. In a flocculated bed, syneresis, as defined here, will occur within the flocs, but not between flocs. Isotropic shrinkage over large distances is not expected. Without gravity they would shrink into themselves, and away from their neighbours. Under gravity they will densify and settle without significant lateral movement. Some shear is generally required to cause densification and syneresis or at least accelerate these processes. Even in the absence of an applied shear through raking, hydrodynamic shear flow is induced by the settling of particles and aggregates of particles while the liquid flowrate is zero on surfaces such as walls and rods. Consequently there is a shear gradient or effective shear rate in the region of these surfaces. This shear rate, though much lower than in the presence of raking, does contribute to densification and syneresis. Both shear densification and syneresis would generate a gap between the settling bed and the cylinder wall. However, if a smooth rod is placed within the bed, the effects should be diametrically opposed. Shear will cause the top of the bed to move away from the rod just as it does at the wall. Syneresis will cause the “hole” generated by the rod to shrink isotropically, as shown in Fig. 1. The gel around the rod will be under compaction due to this shrinkage, and will be pressed against the rod with some force. The second topic to be addressed is the distance to which wall effects extend into the settled bed. If the wall is supporting the bed in contact with it, the bed adjacent to the wall will in turn support the bed further in and so on. In the extreme case an arch would form across the cylinder between opposite walls, and no sedimentation would occur. This is analogous to the arch formed in hard rock tunnels, or the Gothic arch in cathedrals, Fig. 3. A formula has been developed to describe this arching which has a characteristic diameter that depends on the ratio between the shear yield stress and the effective density of the solids. A review is presented on the concepts of syneresis, shear densification and techniques that seek to improve the dewatering process. By considering the cases of flocculated and unflocculated alumina suspensions (AKP-30), these phenomena are investigated in this work by performing settling tests with rods inserted in cylinders. The settling data is analysed to confirm that these phenomena produce significant effects for settling around rods and determine the conditions under which this could have an industrial application. Additionally, dye techniques are used to examine if arching occurs due to wall effects in the settled bed. 1.1. Syneresis Syneresis is based on the isotropic shrinking of a solid to decrease its energy state [10], while shear densification is aggregate contraction or shrinkage due to a shear force [1]. Research on syneresis has been predominantly on organic gels, in particular polymer gels or natural gelled products as are often utilised in the dairy industry. Syneresis is often defined as the expulsion of a liquid as a gel contracts resulting in uniform shrinkage due to the immobilisation of particles [9,11]. Vliet et al. (1991) observed syneresis in suspensions of colloidal particles formed from casein protein molecules [9]. They divided syneresis into two components, one due to external pressure (gravity in this instance) and another termed endogenous syneresis. They focus on the latter and state that large pores in the gel facilitate syneresis and that the extent of syneresis depends on the rate at which the polymer strands yield. Brinker and Scherer (1990) proposed that it occurs because the repulsive double layer that stabilises the sol collapses [10]. They also found that the addition of electrolyte causes additional shrinkage. Scherer (1989) defines syneresis as the contraction of gel spontaneously without solvent evaporation [10]. Scherer studied the syneresis of silica gels, which expel water through the following condensation reaction: Si-OH + HO-Si → Si-O-Si + H2 O. Vysotskii and Strazhesko (1973), along with Raman et al. (1996) found that for silica gels, condensation is the slowest at the isoelectric point and hence syneresis is at a minimum [12,13]. This occurs because, at the isoelectric point, silica gels have the smallest pores making it difficult for the solvent to escape. However, Scherer states that vastly different microstructures can form at other pH values, resulting in varying rates of syneresis due to differences in pore permeability [10]. Scherer suggests it could be caused by the reduction in surface energy and that it can be the result of both the chemical structure and also microstructure of the gel. Raman et al. support this definition, finding that in inorganic gel condensation, reaction rates increase until the interfacial free energies between the Author's personal copy B. Buratto et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 449 (2014) 157–169 solid and liquid becomes unfavourable and that it is an irreversible process [12]. Vachoud and Domard (2001), who studied the syneresis of polymers (chitin) in solvent, state that the release of solvent and resulting decrease in gel volume is due to a thermodynamic imbalance between the solvent and polymer created at the gel point [14]. They found that by manipulating parameters such as temperature and pH, the amount of syneresis could be regulated. Faers et al. (2006), considered syneresis related changes in rheological properties of weak colloidal gels formed by depletion flocculation. They measured the relaxation modulus and their focus was on the migration of polymer in solution relative to that in constrained gels. Helbig et al. (2000), studied the syneresis of alumina particles during pH induced destabilisation (a pH approaching the isoelectric point) of dense colloidal suspensions ( = 0.25 to 0.50) [15]. They found no syneresis effects for solids concentrations,  ≥ 0.35 and proposed that the solidification is not solely a function of minimising energy but that energy and entropy are competing factors which in combination must be minimised in the network structure. This may not correspond to the lowest total energy state and hence shrinkage. 159 1.4. Enhancement of the settling process Densification is the process of releasing trapped water from aggregates when they are disrupted by a shear force [1]. Xin and Lambropoulos (2000) found shear facilitated densification in fused silica and modelled the behaviour using a constitutive law [16]. Selomulya et al. (2003) describe two main densification mechanisms, break-up and deformation [17]. When a shear force is applied that is greater than the floc strength it results in the breakup and densification through re-aggregation. If however, the critical shear force is not met then flocs may only experience a rotating effect but not break the joining bonds, which also results in rearrangement. Densification through rearrangement is preferable, as the break-up mechanism can actually reduce the dewatering ability of suspensions by causing structural damage and making them less permeable [7,8,18,19]. Their research found that restructuring results in denser flocs. Therefore, a low level of applied shear results in a denser network structure. Shear forces are expected to result in shrinkage during densification, as has been shown by Markondeya Raj et al. (2002) in research on tape cast alumina and its shape variation during densification [20]. The effects of shear forces on densification have also been investigated in the processes of granulation and pelletisation. In the granulation process a high shear force is applied to granules resulting in structural changes. Bouwman et al. (2005) found that increasing densification occurred with time in the granulator for particles granulated with pure water [21]. Analysis of the granules indicated that the granule cores underwent increased densification while the outer region was more porous. The rate and extent of dewatering is controlled fundamentally by the permeability and compressibility of a suspension and is dependent on the characteristics of the particulate system involved, such as the particle or aggregate size and shape. However, the extent of dewatering can also be affected by the use of chemical additives, such as flocculants or mechanical devices, such as rakes. Extensive research has been conducted on improving thickener operation through the addition of rakes. Farrow et al. (2000) investigated the consolidation mechanism of flocculated kaolin slurries [24]. They used a mechanical rake at various bed heights to determine the contribution of compression versus shear on dewatering and found that the rake action was the dominant mechanism that caused dewatering through both the densification of aggregates (intra-aggregate liquor) and the removal of inter-aggregate liquor. Usher and Scales (2009) found that settling tests using either rakes and/or conical settling cylinders settled further and at faster rates than for settling in straight sided cylinders [8]. Johnson et al. (2000) [25] demonstrated the importance of shear in compressional dewatering. They found that in a deep bed thickener containing a polymer flocculated alum-based sludge, a shear rate of only 1 rpm, achieved through stirring, settled to 17 wt% solids compared to only 3–4 wt% without the shear effect. A problem that has been identified with the use of shear is that the application of high shear rates for extended times can sometimes reduce the rate and extent of dewatering through the breakage of flocculated aggregates [8,19]. Another feature of the dewatering process that can enhance settling is channelling. In a flocculated suspension which is undergoing thickening, channelling is the formation of flow channels that are of a much larger size than the settling solid particles [22]. Chen et al. (1996) found an enhanced effect which could be attributed to channelling in highly concentrated sludge and suggested that it is due to a spatial imbalance between the inter-floc binding forces and the fluid forces acting on the flocs [23]. As channelling provides a pathway for liquid escape and increases the settling velocity of the particles, it is desirable to promote this effect in thickeners. Thickeners based on promoting channelling have been investigated for the treatment of bauxite residue suspensions [1]. Based on the overview, syneresis and shear densification are well understood concepts; however, they have not been analysed to determine the settling conditions under which each mechanism applies and whether with the addition of rods or other structural elements, either mechanism can be exploited to improve the dewatering process. As such, this work looks to establish whether syneresis or shear densification is the dominant dewatering mechanism in both flocculated and unflocculated alumina suspensions. It looks to analyse the solids interaction with a rod placed in the centre of the cylinder and to analyse settling test data when the rod is included to determine if it affects the settling rate in both the unflocculated and flocculated cases. 1.3. Wall effects 2. Theory The influence of wall effects in the settling process is often ignored, despite the fact that theoretically, walls should have an effect. Zhao (2004) performed settling tests on flocculated water treatment plant sludge and found some evidence for wall effects [22]. It was found that for large flocculant doses, the walls of the cylinder affect the internal networked structure, particularly when the cylinder diameter is small. Chen et al. (1996) showed that when the ratio of the cylinder diameter to the aggregate diameter of the settling agglomerates is greater than between 200 and 1000, wall effects should be negligible [23]. 2.1. Syneresis and shear densification 1.2. Shear densification It is established that syneresis and shear densification are two very different concepts. Syneresis is caused by the rearrangement of particles to reach a lower energy state and results in isotropic shrinkage with expulsion of water. Theoretically therefore, if a rod is placed in the centre of a settling test and the syneresis mechanism is occurring, the solids should shrink away from the walls and towards the rod and cling to it in an isotropic manner, as shown schematically in Fig. 1A. Author's personal copy 160 B. Buratto et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 449 (2014) 157–169 Fig. 1. A) A plan view looking down on a cylinder containing a rod in the centre with two different effects: A) Syneresis resulting in shrinkage away from the cylinder wall and towards the rod, B) shear densification resulting in shrinkage from the rod and the walls. B) Syneresis in a flocculated system. Contact forces between flocs are not large enough to drag the flocs together. The flocs themselves shrink and then continue to settle. The two steps above occur contemporaneously. They are shown separately for clarity. Shear densification on the other hand, is based on the breakdown or rearrangement of loosely agglomerated particles under shear that then reform or form denser structures with the release of water. As a bed starts to form, the solids against the wall are subjected to a shear force, causing densification oriented normal to the shear. Theoretically, if the rod is in the centre of the cylinder the solid should shrink away from both the rod and the walls, as shown schematically in Fig. 1B. 2.2. Settling tests In a settling test, a suspension is made up at a known concentration, 0 , and is poured into a measuring cylinder. As shown in Fig. 2A, initially at zero time (t = 0), the concentration everywhere in the cylinder is 0 . For t > 0 the solids settle under the influence of gravity provided that the solid has a higher density than the liquid. This results in a clear liquid phase (supernatant) and a phase of Fig. 2. Schematic of a batch settling test at different points in time. Author's personal copy B. Buratto et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 449 (2014) 157–169 suspension at the top near the supernatant interface, that is still at concentration o (Fig. 2B). Settling particles eventually connect to form a continuously networked structure that can transmit stress. This occurs at a critical solids concentration known as the gel point, g . At the bottom of the cylinder, the solids concentration will be greater than the gel point as particles settle and form a network that is progressively compressed under the weight of more particles. This is shown in Fig. 2B. Eventually, all of the material forms a continuously networked structure that extends to the solid-liquid interface at the top of the bed, as shown in Fig. 2C [3]. The fully networked suspension consolidates even further and eventually the suspension reaches its final bed height and solids concentration distribution (Fig. 2D). Plotting the height of the interface with time results in a batch settling profile. Batch settling profiles are dependent on the type of suspension being considered. Chen et al. (1996) state that a particulate system typically has four characteristic settling regions [23]. 1) Induction zone – particles change from a state of mixing to settling. 2) Constant rate zone – the interface versus time is a constant function as settling is unhindered resulting in a constant settling velocity. 3) Falling rate zone – a plot of the interface height versus time is curved, with a variable settling velocity generally lower than 161 that in the constant rate section. Particles interact and hinder each other in this zone [26]. 4) Compression zone – the curve flattens out as the final bed height is approached; the solid volume fraction varies with height in the bed. Coe and Clevenger (1916) proposed that the settling rate is a function of local solids concentration, u() [27]. Similarly, Landman and White (1994) defined a hindered settling function, R(), as a correction to Stokes Law for non-zero solids concentrations [28], that provides a measure of the resistance to flow of liquid through the suspension which is related to the settling velocity by 2 R() = g(1 − ) , u() (1) where  is the solid-liquid density difference and g is the standard gravitational acceleration. The solids concentration profile at equilibrium, once settling and consolidation is complete, is determined by the strength of the particle network in compression. This property is also a strong function of local solids concentration and defined as a compressive yield stress, Py (). It increases from zero at the gel point. Batch settling profiles can be analysed to determine these material properties as Py () and R() curve fits using software developed at The University of Melbourne (P527 B-SAMS–Batch Settling Test Analysis and Modelling Software) as described by Usher et al. (2005) [29]. The final height is used to infer Py () and the value of Fig. 3. A) An examples of gothic arches used in architecture, B) Diagram of an arch between two parallel plates and C) Diagram of an arch in a cylinder. Author's personal copy 162 B. Buratto et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 449 (2014) 157–169 the gel point, g , while data in the constant and falling rate zones is used to determine R() data as described by Lester et al. (2005) [3,7,29]. 2.3. Arching To determine the importance of wall effects in the settled bed, an analysis was performed to describe the support of networked (or gelled) solids by cylinder walls. The basis of the analysis is the assumption that the walls support a Gothic arch type structure as shown in Fig. 3A. Initially, this concept was developed to describe the shear forces involved in an arch between two parallel plates (Fig. 3B). The analysis was then extended to an arch in a cylinder (Fig. 3 C), yielding an equation for the maximum diameter, dc , that the arch can span such that dc = 2y , g (2) where  y is the shear yield stress, g is the gravitational acceleration constant and  is the effective density of the suspension. For the alumina suspensions utilised in the experimental phase reported herein, the solid-liquid density difference,  = 3980–1000 = 2980 kg m−3 . The settled bed was measured to have a shear yield stress,  y = 5 Pa at solids volume fraction,  = 0.112. Substituting these values into Eq. (2) gives the maximum diameter of an arch that can be supported, dc = 3.1 mm. This is much smaller than any of the cylinders used or gaps formed, so wall support resulting in arches spanning the cylinders is unlikely to occur and spanning arches are not expected for the test data presented in this work. However, small cavities of up to 3 mm in diameter might be expected to form within the bed. 2.4. Postulates Based on the theory and overview presented, it is hypothesised that: 1) If shear densification occurs during settling then the suspension will settle at a faster rate and settle to a lower final bed height because it will shrink away from the rod, which should aid dewatering. 2) This is based on the concept of channelling increasing the settling velocity. The gap around the rod results in an ‘artificial’ channel through which the water might preferentially escape thus aiding the dewatering process, rather than pushing its way through the porous bed, as is necessary when no rod is present. 3) If syneresis occurs during settling then the suspension will settle more slowly due to the solid shrinking onto and clinging to the rod, hindering the settling process. 4) Syneresis should result in the suspension clinging to the rod. This will increase the drag on the particles and network, slow the rate at which dewatering occurs and prevent the bed settling as fast or as far as the case when this type of hindrance is not present. 5) Arching will occur within the final bed structure, with arches/cavities up to approximately 3 mm in diameter. This is based on the solution to the arching equation and an assumed shear yield stress of 5 Pa. 3. Experimental Apparatus and Procedures 3.1. Materials All experiments were conducted using ceramic grade AKP-30 powder (Alumina), having a density of 3,980 kg.m−3 supplied by Sumitomo Chemical Company, Japan. Suspensions were prepared with a constant background electrolyte concentration using potassium nitrate (Amalar, UK). Samples that were polymer flocculated utilised Magnafloc LT20 (acrylate/acrylamide copolymer) supplied by BASF, Australia. Congo red dye was from Ajax Finchem and red food dye was from Queen Fine Foods. Settling tests were performed in cylinders of different cross sectional areas. These were varied using standard measuring cylinders of different volumes. The measuring cylinder volumes were 250, 500 and 1000 mL with diameters of 40.8, 49.1 and 61.4 mm, respectively. The smaller cylinder was plastic while the larger cylinders were glass. In order to examine the effect of varying the rod and cylinder cross sectional areas, three 400 mm long stainless steel rods of differing diameters were used. Two rods had constant cross sections with 12.7 and 19.1 mm diameters, while the other had a tapered diameter of 6.1 mm at one end, increasing to 19.1 mm over a length of 140 mm and then constant for the remainder of the length as shown in Fig. 4A. In order to flocculate the alumina suspension in situ, a plunger (as shown in Fig. 4B) was used whereby a constant number of plunges and constant plunge rate was utilised. In order to secure the rod in a vertical orientation in the centre of the cylinders after flocculation, secure lids were fabricated with holes the size of the rod diameters (See Fig. 4 C). 3.2. Methods 3.2.1. Slurry preparation Alumina suspensions (AKP-30) were prepared at a solids concentration of 2.91%v/v using 0.01 M potassium nitrate as the solvent. This solids concentration was chosen as it was found to settle significantly within one day and to have finished settling within 2 to 3 days, an appropriate time span. Once prepared, the slurry was sonicated for 10 min and then left on a 500 rpm orbital mixer for 24 h before use. The batches remained on the mixer until just before use. 3.3. Flocculant preparation The Magnafloc LT20 flocculant was prepared as a stock solution (2 g L−1 ) by wetting the polymer with a small amount of ethanol and then making up in water and mixing end over end on a rolling mixer for 24 h. This solution was kept for a maximum of 1 week and was stored at 4 ◦ C and wrapped in foil to avoid light exposure. For use in the settling test, the stock solution was diluted with 0.01 M potassium nitrate to 0.01%w/w LT20 and mixed for 1 h on the rolling mixer before use. Complete polymer relaxation may require weeks, but is of little relevance to industrial practice [30]. Optimum flocculant activity has been observed with stock solution age of order 48–72 h, with a reduction in activity at longer times and at least 24 h required for good flocculant activity for such high molecular weight polymers [30,31]. 3.4. Settling tests Numerous settling test scenarios were conducted including i) no rod, ii) with rod, iii) no rod flocculated and iv) with rod flocculated. For all tests, the volume of slurry, rod and flocculant solution were calculated in advance to ensure that the combined initial height was always 20 cm and when applicable, the flocculant dose was 0.040 g kg−1 (40 g tonne−1 ) LT20. To prepare the flocculated suspension, a measured volume of alumina slurry (2.91%v/v AKP30) was removed from the mechanical mixer and was added to the selected measuring cylinder. Flocculant solution (0.01%w/w LT20) was added and then the plunger shown in Fig. 4B was moved up and down in the cylinder four times at a constant rate to mix the flocculant and slurry over 16 s. A rod was inserted with the plastic Author's personal copy B. Buratto et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 449 (2014) 157–169 163 Fig. 4. A) Dimensions of the tapered rod used in the experiments. B) The plunger used to flocculate the suspensions. C) Schematic diagram of the experimental setup with a plastic lid holding the rod in place. lid to hold it in place. The height of the interface was recorded with time. As the solids settled the level of the suspension-supernatant interface was recorded at regular height intervals. Once the interface began to move slowly, its height was measured with time (e.g. once per hour). The solution was considered to have reached its final height once the height remained the same after a doubling of time. 3.5. Dye experiments To create a dyed solution, the same preparation procedure as for the undyed slurry was used, except when diluting the required amount of concentrated potassium nitrate stock solution to 0.01 M, 1 mL of red food dye was added in place of 1 mL of the water (before adding in to the AKP-30 powder). The settling tests were completed as before. To create a dyed solid, the slurry was prepared in the same way as for the undyed slurry except that before sonicating, 0.2 g of congo red dye was added to the mixture. This was shaken a few times to mix the dye. The settling tests were performed as before. To examine wall effects, at the completion of settling, a 10 mL pipette and syringe set-up was used to excavate the bed away from the walls. The pipette was carefully inserted to just above the top of the bed. The pipette was then carefully lowered in a vertical motion 10 mm into the bed as the syringe was slowly drawn, removing part of the bed. This was repeated until a clear view of the internal structure of the bed could be seen and observations recorded. Table 1 Settling testwork programme. # Cylinder Volume (mL) Flocculated Rod Diameter (mm) 1. 2. 3. 4. 5. 6. 7. 500 500 500 250, 500 and 1000 500 250, 500 and 1000 250, 500 and 1000 Yes Yes Yes Yes 12.7 mm 19.1 mm 12.7 mm 19.1 mm Tapered of syneresis and shear densification were recorded/photographed. Some settling tests were conducted with dyed liquid while in others, the solids were dyed and then the settled equilibrium bed excavated. 4. Results 4.1. Flocculated settling Fig. 5 presents data for the interface height versus time of settling for a 2.78%v/v AKP-30 slurry with 0.01 M potassium nitrate electrolyte and 40 g t−1 LT20 flocculant. The figure shows the settling curves for the cases of no rod, the constant cross section rod (diameter 19.12 mm) and the tapered rod. With each graph representing data for a different cylinder size, A) 250 mL, B) 500 mL and C) 1000 mL. The following trends were observed: 3.6. Work programme The variables considered in the settling tests were: flocculated versus unflocculated/coagulated suspensions, rod diameter, measuring cylinder diameter and rod type (tapered or constant cross section). The tests conducted are summarised in Table 1. While the experiments were being conducted, observations indicative • In all cases, irrespective of the cylinder size, when there was a rod, whether tapered or of constant cross section, settling was quicker than in the case with no rod. • The tapered rod had a very similar settling profile to that of the constant cross section rod, although it always finished at the lowest final bed height compared to the constant rod and no rod Author's personal copy 164 B. Buratto et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 449 (2014) 157–169 0.25 0.25 No Rod Rod (d = 19.12mm) No Rod Rod (d = 19.12mm) Tapered Rod 0.15 0.1 Tapered 0.2 Height (m) Height (m) 0.2 0.15 0.1 0.05 0.05 A: 250 mL B: 500 mL 0 0 1 10 100 1000 10000 1 100000 10 100 1000 10000 0.25 0.25 No Rod No Rod Rod (d= 12.70mm) Rod (d = 19.12mm) Tapered Rod 0.2 0.15 0.2 Height (m) Height (m) 100000 1000000 Time (s) Time (s) 0.1 0.05 0.15 0.1 0.05 C: 1000 mL D: 500 mL 0 0 1 10 100 1000 10000 100000 1 10 Time (s) 100 1000 10000 100000 Time (s) Fig. 5. Height versus time plots for a 2.78%v/v AKP30, pH 7.81 with LT20 flocculant, for the cases with no rod, the 19.12 mm diameter constant cross section rod and the tapered rod: A) 250 mL cylinder, B) 500 mL cylinder and C) 1000 mL cylinder, D) shows the settling curve for the 12.70 mm diameter constant cross section rod in a 500 mL cylinder. cases. However, this is expected due to the reduction in rod cross section at lower heights. 0.25 Rod (d=19.12mm), 250mL Rod (d=19.12mm), 500mL 0.2 Rod (d=19.12mm), 1000mL No Rod, 250mL Height, h (t (m) Fig. 5D shows the settling curve data for the thinner diameter rod (12.70 mm) in the 500 mL measuring cylinder, with the same general trends as for the 19.12 mm rod case. When comparing data, it should be noted that there are slight variations due to different batches of slurry being used and the experiments being completed at different times. Fig. 6 presents a comparison of the settling curves for the cases of constant diameter rod (19.12 mm) and no rod, for the 3 different cylinder volumes. The curves for the cases with the rod are all in close agreement, while the cases without the rod are in close agreement, except for the short time behaviour in the 1000 mL cylinder. A clear distinction can be made between settling with rods and without rods. Using proprietary software, the hindered settling functions for the settling data for the 19.12 mm diameter rod were determined and are presented in Fig. 7. The settling tests conducted with the rod are below those without the rod, indicating the particles were less hindered during settling. Fig. 8 presents data of the final settled height for the cases of a constant cross section rod (diameter 19.12 mm), no rod and tapered rod for the three different size cylinders. The following observations were noted: No Rod, 500mL 0.15 No Rod, 1000mL 0.1 0.05 0 10 100 1000 10000 100000 Time, t (s) Fig. 6. Height versus time plot comparing the cases of constant rod cross section (diameter 19.12 mm) and no rod for 250 mL, 500 mL and 1000 mL measuring cylinders for the 2.78% v/v AKP-30 suspension. Author's personal copy -3 Rod (d=19.12mm), 250mL Rod (d=19.12mm), 500mL Rod No Rod Tapered Rod 0.058 Rod (d=19.12mm), 1000mL 0.056 No Rod, 250mL 1.E+10 No Rod, 500mL Final Height (m) Hindered Settling Function,R ( 165 0.06 1.E+11 -1 (kg s m ) B. Buratto et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 449 (2014) 157–169 No Rod, 1000mL 1.E+09 0.054 0.052 0.05 0.048 0.046 1.E+08 0.044 0.042 0.04 1.E+07 0 0.02 0.04 0.06 Solids Volume Fraction, 0.08 0 0.1 400 600 800 1000 1200 Volume (mL) Fig. 7. Hindered settling function, R() (kg s-1 m-3 ) versus solids volume fraction,  (v/v) plot comparing the cases of constant rod cross section (diameter 19.12) and no rod for 250 mL, 500 mL and 1000 mL measuring cylinders for the 2.78% v/v AKP-30 suspension. Fig. 8. Final height versus volume for the constant cross section rod (diameter 19.12 mm), no rod and tapered rod cases, for the 3 different volume cylinders, 250 mL, 500 mL and 1000 mL. improvement in settling flux. Also, thickener performance would be enhanced by more than just the increased settling rate, with the potential extent of dewatering also improved for these densified aggregates. • When there is no rod, variation of the final settling height with cylinder diameter is observed to be effectively constant and less significant than observed random experimental variation. • The presence of the constant cross section rod significantly affects the final height of the solid bed, the final height decreases with increasing cylinder size (slope of the line is -9 × 10−6 m mL−1 ). • The tapered rod has the lowest final height in all cylinder sizes and decreases slightly with volume of the cylinder (slope of the line is -3 × 10−6 m mL−1 ), mainly due to its shape. 4.2. Unflocculated/coagulated settling Fig. 9 shows the settling test results for unflocculated/coagulated slurries with no rod, the 19.12 mm diameter rod and the 12.70 mm diameter rod in 500 mL cylinders. The following trends were observed: These trends indicate that the cross section of the rod relative to that of the cylinder is important. The 250 mL cylinder results in a hold up of the solid bed preventing it from settling as far, outweighing the channelling benefit of the rod. In the 1000 mL cylinder the rod does not cause a hold up effect and the bed can settle further, possibly due to shear densification. In the tests conducted, the reduction in area due to the rod ranges from 4–22%, while the improvement in settling rate was up to 200% for flocculated suspensions. Overall, there is an expected • For both diameter rods, when the rod is included, the suspension settles more slowly than when the rod is absent. • The effect of the rod slowing down the settling appears to be more significant for the large diameter rod. • In both cases the rod and no rod curves cross over in the compression region, however, the final heights for both are not significantly different. 0.25 0.25 Rod (d = 19.12mm) Rod (d = 12.70mm) No Rod No Rod 0.2 0.2 Height (m) Height (m) 200 (v/v) 0.15 0.1 0.15 0.1 0.05 0.05 A B 0 0 1 10 100 1000 Time (s) 10000 100000 1 10 100 1000 10000 100000 100000 Time (seconds) Fig. 9. Height versus time plot for a 2.91%v/v AKP30, pH 7.81 without flocculant settling in a 500mL measuring cylinder for the cases with no rod and with rod diameters: A) 19.12 mm and B) 12.70 m. Author's personal copy B. Buratto et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 449 (2014) 157–169 1.E+12 -3 (kg s m ) 166 Hindered Settling Function,R ( -1 Rod (d = 19.12mm) No Rod 1.E+11 1.E+10 1.E+09 1.E+08 500 mL 1.E+07 0 0.02 0.04 0.06 Solids Volume Fraction, 0.08 0.1 (v/v) Fig. 10. Hindered settling function, R(f) (kg s-1 m-3) versus solids volume fraction, f (v/v) plot comparing the cases of no rod with a rod diameter of 19.12 mm in a 500 mL measuring cylinder. The settling curves show the reverse situation compared to the flocculated suspension, with the rod hindering the settling compared with the no rod situation. In all cases the final heights are similar. The curves do not follow as closely as in the case of flocculated settling. It would be expected that the ‘no rod’ cases in Fig. 9 should be identical because the tests were the same. The only difference is that two different batches of slurry were used. The error is likely to be due to the slurry preparation differences and pH drift. In the experiment for the thinner rod the slurry was not sonicated unlike the batch for the thicker rod (sonication was introduced to improve the homogeneity of the slurries). However, in both cases the general trend is the same, namely the rod appears to hinder settling. Fig. 10 shows the hindered settling curves for the 19.12 mm diameter rod for the cases of rod and no rod; the rod clearly hinders settling. These experiments are not always straight-forward and a number of factors influence the results. In particular, there are issues with recording uneven interfaces, reproducibility of inserting the rod in the suspension, flocculating reproducibly using the plunger method, keeping the solids concentration constant and maintaining the pH accurately. The latter is important to the non-flocculated case where a constant state of aggregation is critically dependent on the pH. For unflocculated samples, reproducibility was established Fig. 11. Images of flocculated settling tests; A) with constant cross section rod: slurry bed is displaced from the rod leaving an observable gap. B) with the tapered rod slurry bed displaced from the rod leaving an observable gap and C) with no rod, but with a gap of height of approximately 5 mm between the top of the bed and the wall of the cylinder. Fig. 12. Images of a coagulated suspension settling test: A) with a constant cross section rod, the solids cling to the rod, with no gap B) constant cross section rod and C) tapered rod. Author's personal copy B. Buratto et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 449 (2014) 157–169 167 Fig. 13. The final bed of a AKP-30 suspensions; A) unflocculated/coagulated, where the solids are dyed with congo red, B) flocculated, where the water has been dyed with red food dye and C) unflocculated/coagulated, where dyed liquid has been allowed to diffuse into the bed crevices. by repeating the settling of some samples, with excellent results that could be superimposed. For flocculated samples, tests could not be repeated without preparing a new batch of suspension. However, wall effects were much more significant for flocculated samples and though sedimentation was more variable, trends in behaviour could be clearly identified. Observed trends relating to effects of syneresis and densification are consistent across all test data and subsequent analysis outputs. Due to the reproducibility of the trends, experimental uncertainties are not expected to have had a significant influence on the implications of the results presented. 5. Discussion 5.1. Shear densification and syneresis The settling test results indicate that when the alumina suspension is flocculated with 40 g t−1 of LT20, the addition of a rod, whether constant cross section or tapered, results in the suspension settling at a faster rate than when the rod is absent. The final bed height for the constant cross section rod is affected by the size of the cylinder; in a 1000 mL cylinder it appears to settle further than the no rod case. The tapered rod however, mainly due to its shape, always had the lowest final bed height. When the slurry was only coagulated but not flocculated, the rod appeared to hinder the settling of the solids resulting in their settling at a slower rate compared to the no-rod situation. No significant difference could be seen in the final bed heights for the unflocculated/coagulated suspensions. It is proposed herein that this difference is due to the different mechanisms, shear densification and syneresis, occurring in the flocculated and coagulated suspensions respectively. Fig. 11 shows images for the flocculated case where as settling progressed and a bed started to form, the solids moved away from the constant cross section rod, leaving an observable gap around its perimeter. When the rod was removed, the rod was completely clean, with no solids attached to it. This phenomenon was enhanced for the case of the tapered rod with a larger observable gap and again the rod was clean. When the constant rod was removed the bed collapsed, whereas, the bed remained intact when the tapered rod was removed. Additionally, a 5 mm gap was noted at the top of the bed between the measuring cylinder walls and the solid bed Author's personal copy 168 B. Buratto et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 449 (2014) 157–169 for the case without the rod. Note that, during regular settling as the material settles to form a networked bed, solid particles are forced into close contact with the outer wall, where the presence of a shear yield stress will hold up the particles near the wall relative to those further away from a surface. The net effect is a reduced settling rate near the outer wall, with a relatively higher settling rate away from the outer wall, which eventually leads to a radial bed height differential that grows until the higher outer bed material undergoes inward collapse to produce a more even bed height. The fact that the solids moved away from the rod and that the gap at the wall occurred in the flocculated case with no rod, indicates that shear densification was likely to be the consolidation mechanism. Fig. 12 shows images for the coagulated suspension case whereby as settling progressed and a bed started to form, particles adhered to the rod, there was no gap between the solids and the wall and there was no difference between the tapered and nontapered rod cases. These results are in contrast to those obtained for the flocculated suspension. The fact that the solids stuck to the rod, but not the wall, and a ball of solids around the base was intact as the rod was lifted up, suggests that the solids shrink towards the rod, indicating syneresis was likely to be the consolidation mechanism. It would seem that the final height is not significantly affected by syneresis. The results support the postulate that shear densification aids the rate of settling while syneresis provides a hindrance. The interesting aspect is the clear difference in behaviour between the flocculated and coagulated suspensions. The role of rods as dewatering aids in industrial settling is highlighted but within the context that they might not be useful in all suspensions. The use of congo red to dye the solids allowed the final settled bed structure to be observed more clearly. Fig. 13A shows the internal structure of a settled unflocculated/coagulated bed, showing cavities filled with water throughout the structure, not solely at the walls. The predicted maximum self-supporting arch size of 3 mm, determined using Eq. (2), is not observed in any of these cavities. The dye only coloured the water and was used to identify regions of low solids density and hence high liquid density, to determine if the network was being supported by the walls. Fig. 13B shows the final bed of a flocculated alumina slurry. The dye highlights definite holes near the wall in the top section of the bed. Compared to the internal cavities, those near the wall are much larger. The magnified photo shows the largest cavity seen in the beds and its diameter is approximately 3.1 mm. This agrees well with the predicted maximum value. Although the derivation was for arching across an entire cylinder, it seems reasonable that within the structure, smaller systems with similar supports exist. Fig. 13 C presents a dyed, unflocculated suspension. In this case, the water was removed down to the surface of the bed, and a layer of red food dye placed on top. With time it gradually seeped into the gaps in the structure and dyed the water. Unlike what was observed for the flocculated case, the holes in the structure are much smaller, but still present. Internally, no large cavities of order 3 mm were observed, however, definite holes can be seen of up to about 1 mm in size, suggesting that arching does occur for the unflocculated/coagulated material as well. The existence of cavities in the final bed structures suggests the shear yield stress of the material reduces its ability to dewater and this is more significant near the walls of the cylinder; indicating that wall effects may become significant in certain circumstances. This is important because arching reduces the extent of dewatering that is achievable. Mechanisms that gently break up these cavities have the potential to significantly aid dewatering and improve the extent of dewatering in gravity thickener operations. 6. Conclusions This research has investigated the phenomena of syneresis and shear densification using comparative settling tests that included cases without a rod, with a rod present, without flocculant but coagulated and with LT20 flocculant. It was found that the shear densification mechanism occurs when the suspension is flocculated and this results in the suspension settling at a faster rate when a rod is present compared to a similar suspension without the rod. This is thought to be due to a channel forming around the rod and shear induced floc densification, both aiding the dewatering process. When the cylinder was greater than 600 mL, the rod resulted in a lower final bed height. When the solution is only coagulated, i.e. not polymer flocculated, the syneresis mechanism occurs and this results in the suspension settling more slowly when a rod is present because the slurry clings to the rod hindering the settling process. This is important from an industrial perspective because when a flocculant is used, the inclusion of rods inside a thickener/dewatering device may increase the rate at which settling occurs and result in a higher solids bed or enable a higher throughput. Such methods may be more favourable than the use of rakes, which are more complex devices and can damage the floc structure. Additionally, creating channels using rods should be more reliable than trying to promote ‘natural’ channelling in thickeners. Dye tests support the concept of internal arches within the bed. Cavity diameters of up to 3.1 mm for the alumina suspensions were observed near the cylinder walls. This agrees well with the theoretical value. These arches are likely to hamper the extent of dewatering, suggesting that the cylinder walls can have some effect on the final bed height. Acknowledgements The authors wish to acknowledge the financial support of the Particulate Fluids Processing Centre a Special Research Centre of the Australian Research Council. References [1] R.G. Holdich, G. Butt, Compression and channelling in gravity sedimenting systems, Minerals Engineering 9 (1996) 115–131. [2] G.J. Kynch, A theory of sedimentation, Journal of the Chemical Society: Faraday Transactions 48 (1952) 166–176. [3] D.R. Lester, S.P. Usher, P.J. Scales, Estimation of the hindered settling function R() from batch-settling tests, AIChE J. 51 (2005) 1158–1168. [4] J. Hulston, Effect of flocculation conditions on the dewaterability of hematite and red mud suspensions PhD Thesis in: Department of Chemical and Biomolecular Engineering, The University of Melbourne, Melbourne, Australia, 2002, pp. 284. [5] J. Hulston, R.G. de Kretser, P.J. Scales, Effects of flocculation conditions on the dewaterability of hematite and red mud suspensions in: 6th International, Alumina Quality Workshop, AQW Inc, Brisbane, Australia, 2002, pp. 294–300. [6] J. Hulston, P.J. Scales, Dewatering red mud - The effects of flocculation, in: Chemical Engineering in Australia (2001) 17–21. [7] B.B.G. van Deventer, S.P. Usher, A. Kumar, M. Rudman, P.J. Scales, Aggregate densification and batch settling, Chemical Engineering Journal 171 (2011) 141–151. [8] S.P. Usher, P.J. Scales, Predicting settler/clarifier behaviour: The role of shear effects, Filtration 9 (2009) 308–314. [9] T.V. van Vliet, H.J.M. van Dijk, P. Zoon, P. Walstra, Relation between syneresis and rheological properties of particle gels, Colloid Polymer Science 269 (1991) 620–627. [10] G.W. Scherer, Mechanics of syneresis: I. Theory, Journal of Non-Crystalline Solids 108 (1989) 18–27. [11] D.S. Horn, G.L. Messing, Alumina monolith formation by flocculation of boehmite sols, Journal of the American Ceramic Society 72 (1989) 1719. [12] N.K. Raman, M.T. Anderson, C.J. Brinker, Template-based approaches to the preparation of amorphous, nanoporous silicas, Chem, Mater. 8 (1996) 1682–1701. [13] Z.Z. Vysotskii, D.N. Strazhesko, Adsorption and Adsorbents, Wiley, New York, 1973, pp. 55. [14] L. Vachoud, A. Domard, Physicochemical properties of physical chitin hydrogels: Modeling and relation with the mechanical properties, Biomacromolecules 2 (2001) 1294–1300. Author's personal copy B. Buratto et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 449 (2014) 157–169 [15] J.M. Helbig, M. Hütter, U.P. Schönholzer, Lack of Syneresis during Gelation of Dense Colloidal Suspensions, Journal of Colloid and Interface Science 222 (2000) 46–50. [16] K.J.C. Xin, Lambropoulos, Densification of fused silica: effects on nanoindentation, Inorganic Optical Materials II 41 (2000) 112–121. [17] C. Selomulya, G. Bushell, R. Amal, T.D. Waite, Understanding the role of restructuring in flocculation: The application of a population balance model, Chemical Engineering Science 58 (2003) 327–338. [18] S.P. Usher, R. Spehar, A. Kiviti-Manor, P.J. Scales, Thickener modelling – Incorporating shear effects, in: Proceedings of XXV International Mineral Processing Congress (IMPC 2010), Brisbane, Australia (2010) 3975–3982. [19] S.P. Usher, R. Spehar, P.J. Scales, Shear effects in thickening, in: R. Jewell, A. Fourie (Eds.) Proceedings of the 13th International Seminar on Paste and Thickened Tailings, PASTE 2010, Australian Centre for Geomechanics, Toronto, Canada (2010) 375–384. [20] P. Markondeya Raj, A. Odulena, W.R. Cannon, Anisotropic shrinkage during sintering of particle-oriented systems-numerical simulation and experimental studies, Acta Materialia 50 (2002) 2559–2570. [21] A.M. Bouwman, M.J. Henstra, D. Westerman, J.T.M. Chung, Z. Zhang, A. Ingram, J.P.K. Seviller, H.W. Frijlink, The effect of the amount of binder liquid on the granulation mechanisms and structure of microcrystalline cellulose granules prepared by high shear granulation, International Journal of Pharmaceutics 290 (2005) 129–136. [22] Y.Q. Zhao, Settling behaviour of polymer flocculated water-treatment sludge I: Analyses of settling curves, Separation and Purification Technology 35 (2004) 71–80. View publication stats 169 [23] G.W. Chen, I.L. Chang, W.T. Hung, D.J. Lee, Regimes for zone settling of waste activated sludges, Water Research 30 (1996) 1844–1850. [24] J.B. Farrow, R.R.M. Johnston, K. Simic, J.D. Swift, Consolidation and aggregate densification during gravity thickening, Chem. Eng. J. 80 (2000) 141– 148. [25] S.B. Johnson, P.J. Scales, D.R. Dixon, M. Pascoe, Use of a superthickener device to concentrate potable water sludge, Water Research 34 (2000) 288– 294. [26] S.C.A. França, G. Massarani, E.C.J. Biscaia, Study on batch sedimentation simulation - establishment of constitutive equations, Powder Technology 101 (1999) 157–164. [27] H.S. Coe, G.H. Clevenger, Methods for determining the capacities of slimesettling tanks, AIME Trans. 55 (1916) 356–384. [28] K.A. Landman, L.R. White, Solid/liquid separation of flocculated suspensions, Advances in Colloid and Interface Science 51 (1994) 175–246. [29] S.P. Usher, B. Gladman, P.J. Scales, New techniques for understanding settler/clarifier behaviour, in: American Filtration and Separations Society, 18th Annual Conference, American Filtration and Separations Society, Atlanta GA USA, April 10-13 2005 pp. CD Proceedings (2005). [30] A.T. Owen, P.D. Fawell, J.D. Swift, J.B. Farrow, The impact of polyacrylamide flocculant solution age on flocculation performance, International Journal of Mineral Processing 67 (2002) 123–144. [31] A.T. Owen, P.D. Fawell, J.D. Swift, The preparation and ageing of acrylamide/acrylate copolymer flocculant solutions, International Journal of Mineral Processing 84 (2007) 3–14.