Hydrodynamics of deinking flotation

Int. J. Miner. Process. 56 Ž1999. 277–316

Hydrodynamics of deinking flotation Franc¸ois Julien Saint Amand

)

Centre Technique du Papier Domaine UniÕersitaire, B.P. 251 38044, Grenoble, Cedex 9, France Received 19 January 1998; received in revised form 17 August 1998; accepted 27 November 1998

Abstract The technological parameters relating to the aeration, mixing and separation steps of the flotation cells currently used in the field of waste paper recycling are reviewed in this paper. A laboratory flotation cell was designed to study the effects of the hydrodynamic flotation parameters, and tested on different deinking furnish. The physico-chemical aspects of flotation deinking were not investigated and the flotation efficiencies were measured for the ink particles detached from the fiber surface. Experimental evidence concerning the first-order kinetics of ink removal, with respect to flotation time and consumed air, was obtained for different particle and bubble sizes. Small bubbles were shown to be more effective, but led to higher fiber losses than large bubbles for a given ratio of consumed air. The design of the experimental cell was then improved in such a way as to produce calibrated air bubbles of different sizes between 0.5 and 2 mm diameter, and to investigate separately the effect of turbulence generated in the flotation cell. The results showed that the relation between the flotation rate constant and bubble size depends on particle size and shape. The optimum ink particle size showed to be in the range of 10–100 mm. The flotation efficiencies decreased strongly with particle size in the case of flat shaped ink particles, while no significant efficiency drop was observed for the large particles in the case of calibrated laser inks. A turbulence increase improved significantly the removal efficiency except for the largest laser ink particles. The experimental results were compared with the theory developed in the field of mineral flotation and showed quite good agreement with the theoretical relations as far as the effects of air ratio, bubble size and turbulence with respect to relatively small particle sizes are concerned. q 1999 Elsevier Science B.V. All rights reserved. Keywords: flotation; waste paper; deinking; kinetics; bubbles; turbulence

)

Corresponding author. Fax: q33-4-76-15-40-16

0301-7516r99r$ - see front matter q 1999 Elsevier Science B.V. All rights reserved. PII: S 0 3 0 1 - 7 5 1 6 Ž 9 8 . 0 0 0 5 0 - 7

278

F. Julien Saint Amandr Int. J. Miner. Process. 56 (1999) 277–316

1. Introduction Froth flotation technology has been developed and used for many decades in the mineral processing industry before the technology was transferred to the pulp and paper industry for the deinking of waste papers at the beginning of the 1960s. The growth of installed flotation units has then been relentless in the field of deinking. Today, in Europe as well as in the North America and other countries, flotation deinking is the most significant stage in the waste paper recycling process for the production of graphic grades. Improved flotation conditions in terms of ink and speck removal efficiency and controlled rejects are requested to fulfill the brightness and cleanliness requirements of high quality paper products. In order to meet these requirements under economical conditions, the subsequent flotation steps of collision and attachment between ink particles and air bubbles, and the selective removal of the particle-bubble aggregates have to be optimized. Among these individual microprocesses, the mechanisms of attachment of ink particle onto the air bubbles are mainly driven by physico-chemical interactions, while the collision and removal of the inked bubbles processes are driven by hydrodynamics. Extensive theoretical and experimental studies on the various aspects of froth flotation, including fluid dynamics, surface chemistry and broadly speaking colloid science, were carried out for the mining industry ŽIves, 1984; Laskowski, 1989.. However, many of the fundamentals are believed to be applicable to deinking flotation because of major similarities. The most significant differences between deinking flotation and mineral flotation are to be found in the development of specific equipment as illustrated in Appendix A, and especially in the particular characteristics of waste paper pulp suspensions including the following. . The wide distribution of size, shape and surface properties of the particles to be removed: mainly ink particles, from about 1 mm to 1 mm, generally hydrophobic except for water-based inks, and flat shaped for the large particles, since other techniques such as screening with slots down to 0.1 mm and centrifugal cleaning are also used to remove the various impurities of waste paper pulp suspensions Žpressure sensitive adhesives, hot melt glues, plastic films, printing inks, . . . .. . The low density of the particles to be removed from the deinking pulp: polymeric particle with specific gravity close to that of the water Žmineral particles used as fillers and for paper coating, such as kaolin and calcium carbonate in the size range of 1 mm, should generally not be removed.. . The trend of cellulose fibers Žtypically 1–3 mm in length, and 10–30 mm in diameter, depending on the wood species. to flocculate Žform flocs or a network. at consistencies above about 1%, as soon as the turbulence level is decreased as observed in the separation zone of deinking cells. . The chemicals added to the repulped waste papers to release the ink particles from the fibers and to enhance the flotation process Žcalcium soap and caustic soda or other deinking chemicals to be used under alkaline or neutral conditions., as well as the various chemicals introduced with the waste papers Žespecially surfactants used in the coating color..

F. Julien Saint Amandr Int. J. Miner. Process. 56 (1999) 277–316

279

During the period that the flotation technique has been employed in deinking, the basic investigations were first focused on physical chemistry, because of the very complex properties of recycled fiber suspensions in terms of dissolved, suspended and colloidal components as well as added deinking chemicals and surfactants from the recovered papers ŽSchweizer, 1965; Bechstein and Unger, 1972; Ortner et al., 1975; Galland et al., 1977; Fischer, 1982; Hornfeck, 1982; Larson et al., 1982; Putz et al., 1991; Stratton, 1992; Hou and Hui, 1993; Santos et al., 1996.. Several theories have been proposed to explain the mechanisms of ink collection onto air bubbles. Investigations on the hydrodynamics of flotation deinking were carried out more recently. The research work reported concerned theoretical studies based on the scientific knowledge gained from mineral flotation, as well as experimental studies carried out on laboratory equipment ŽSchulze, 1991; Julien Saint Amand and Perrin, 1991; Schulze, 1994; Pan et al., 1992; Paulsen et al., 1993; Pan et al., 1993, 1994; Klein et al., 1994; Gottsching et al., 1995; Dorris and Page, 1995; Schmidt and Berg, 1996; Vidotti ¨ et al., 1995; Kocer et al., 1995.. Flotation hydrodynamics governing the aeration, collection and separation steps of the flotation process have also been investigated by the equipment suppliers to develop new technologies. The technical characteristics of current flotation equipment have been described in numerous publications ŽBarnscheidt, 1985; Barnscheidt, 1987; Pfalzer and Schweiss, 1988; Mac Cool and Carroll, 1990; Ferguson, 1991; Gilkey and Yoshida, 1992; Kurz and Frymark, 1992; Torregrossa et al., 1992; Britz, 1993; Britz and Peschl, 1994; Gilkey et al., 1994; Linck and Siewert, 1994; Dessureault et al., 1995; Chudacek et al., 1995; Floccia, 1995; Mac Kinney, 1996; Carlton, 1996; Serres and Colin, 1996; Carletti and Wood, 1996; Hori, 1996., but no detailed data are available on relevant hydrodynamic parameters such as the turbulence level and the bubble size distribution. A review of current deinking flotation technology is given in Appendix A. The principle and design of the different types of deinking cells are compared with respect to the characteristics of the aeration, mixing and separation steps of the flotation process. The analysis of these data was used to establish relevant hydrodynamic flotation conditions to be investigated experimentally. 2. Flotation kinetics It is generally agreed that flotation is a first order kinetics process with respect to the number of particles and bubbles of fixed characteristics, which interact per unit volume and time. This has been extensively discussed and shown in the field of mineral flotation ŽIves, 1984; Laskowski, 1989. as well as more recently in the case of various ink particles in fiber suspensions ŽJulien Saint Amand and Perrin, 1991; Pan et al., 1994; Dorris and Page, 1995.. The simplest form of kinetic equation describing the flotation microprocesses involved in particle bubble collection is given by the variation with flotation time of the number of particles in a given cell volume ŽSchulze, 1994., where particles collected are effectively removed from this volume: d Nprdt s yzNp Nb Pc Pa Ps Ž 1.

280

F. Julien Saint Amandr Int. J. Miner. Process. 56 (1999) 277–316

where Np and Nb are the numbers of particles and bubbles per unit volume, Z s zNp Nb is the number of particle bubble collisions per unit time and volume, and Pc , Pa and Ps are the probabilities of particle bubble collision, attachment and stabilization against external stress forces. Flotation rate constants usually refer to flotation time Ž k s zNb Pc Pa Ps from Eq. Ž1.. which is convenient for comparing the relative floatability of different materials under a given set of conditions in laboratory cells with internal recirculation and aeration, but does not allow different types of laboratory or industrial cells to be compared. Such comparisons are much easier if the volume air to pulp ratio is considered instead of the flotation time to define the flotation rate constants. If one considers a small relative variation D NprNp of particle number in a unit volume of aerated pulp during the contact time D t s t c under given aeration conditions Žair ratio and bubble size distribution. and turbulence, assuming that the number D Np of particles collected during the contact time t c is proportional to the total number Np of particles in the unit volume Žfirst order kinetics. Eq. Ž1. can be written in the form: D NprDt s D Nprtc s yz Np Nb Pc Pa Ps s ykNp

Ž 2.

A similar equation can be written with respect to the air ratio T and the associated flotation rate constant K, if one considers a small fraction DT of the air ratio: D NprDt s yKNp or D NprNp s yKDT

Ž 3.

where DT is a function of bubble diameter d b and number Nb per unit volume: DT s Nb Ž p d b3r6 . which, from Eq. Ž2. gives: D NprDt s yz Np Ž 6 t crp d b3 . Pc Pa Ps or D NprNp s yzPc Pa Ps Ž 6rp d b3 . t c DT Ž 4 . The flotation rate constant K with respect to the air ratio is then given by: K s z Ž 6 t crp d b3 . Pc Pa Ps

Ž 5.

For small variations of D NprNp and DT and after integration, Eq. Ž3. takes the following form where Npo is the particle concentration at T s 0: ln NprNpo s yKT

Ž 6.

The flotation efficiency E for a given air ratio is then defined by: E s 1 y NprNpo s 1 y exp Ž yKT .

Ž 7.

In reality, Eqs. Ž2. – Ž7. are not rigorous to describe real flotation conditions since it is assumed that the successive collection processes at each aeration step Ž DT increase of the air ratio. are achieved under constant conditions Žair amount, bubble size distribution and turbulence. and over a very short time with respect to the proportion of collected particles, which is generally not the case Žtypically about 50% air and decreasing turbulence at each aeration step of industrial units.. However, the approximation is convenient as most of the flotation installations are based either on several aeration steps in the case of continuous industrial flotation

F. Julien Saint Amandr Int. J. Miner. Process. 56 (1999) 277–316

281

installations, or on recirculation in the case of laboratory batch flotation devices. In the specific case of column flotation, the successive collection steps are achieved in one aeration step under roughly constant conditions and over the cell height which defines the contact time. The influence of contact time Ž t c . and air ratio Ž DT . is given in Eq. Ž4., which shows that, the particle flotation rate is proportional to the residence time and amount of air introduced in the cell. The use of Eqs. Ž6. and Ž7. to calculate flotation rate constants with respect to the air ratio, and to assess flotation efficiencies is illustrated in Figs. 1 and 2. The results in Fig. 1 were obtained in the frame of laboratory flotation studies which first showed experimental evidence concerning the first-order kinetics of the deinking process ŽJulien Saint Amand and Perrin, 1991.. Deinking furnish containing small microscopic ink particles from magazines and large flat shaped specks Žthickness: about 5 mm. from UV varnished prints was tested on the first experimental cell ŽFig. 3. operated to produce bubbles in the size range of 2–3 mm. The large ink particles gave the highest flotation rate constant Ž1.2 for 30-mm particles., and a large decrease of the rate constant with particle size was observed for the specks Žfrom about 0.7 down to 0.2 for 300–800 mm particles.. The relations between flotation rate constant and efficiency are shown in Fig. 2. A rate constant of about 1 is necessary to achieve high particle removal efficiency up to 95% for about 300% air ratio which is a common figure for industrial installations Žsee Table 3 in Appendix A.. A rate constant lower than 0.2 corresponds to very low floatability while a rate constant higher than 5 ensures almost complete particle removal. Good correlation should be obtained between laboratory and industrial flotation conditions for similar distributions of bubble size, turbulence level and contact time, as

Fig. 1. Flotation results obtained on the first experimental cell Žbubble size: 2–3 mm. in the case of repulped magazines containing specks from UV varnished prints.

282

F. Julien Saint Amandr Int. J. Miner. Process. 56 (1999) 277–316

Fig. 2. Theoretical flotation efficiency E vs. air ratio T and flotation rate constant K.

far as the inked particles are correctly removed with the foam. In addition it has to be pointed out that only the ratio of removed air should be considered since small bubbles which may be dragged in the accepts do not contribute to the particle removal efficiency. The smallest bubbles do not contribute significantly to the air ratio whereas

Fig. 3. Schematic of the two experimental flotation cells designed to study the effect of bubble size and turbulence.

F. Julien Saint Amandr Int. J. Miner. Process. 56 (1999) 277–316

283

relatively few large bubbles may increase significantly the air ratio, though their contribution to the flotation efficiency is low.

3. Experimental equipment Experimental laboratory scale flotation cells were designed in order to produce variable bubble sizes and turbulence. The design of the experimental cells had to be improved in the course of the study about flotation hydrodynamics, since it proved to be difficult to produce air bubbles in a narrow size range, at different turbulence levels. The main characteristics of the two experimental cells designed to produce variable bubble sizes and turbulence are illustrated in Fig. 3. Both cells were operated in recirculation on a 20-l chest at 10 lrmin feed flow. During the flotation trials, samples were taken from the accepts at different times corresponding to cumulated air-flow rates between 50 and 400%. The first experimental cell was cylindrical Ždiameter: 120 mm. with upward flow Ž1.5 cmrs superficial pulp flow velocity. as shown in Fig. 3. Pulp and air were introduced tangentially at the bottom of the cell which was equipped with a variable speed stirrer. A fixed shaft with radial fins above the stirrer had to be implemented in the cell to avoid vortex formation. Accept pulp was recovered from a peripheral downward zone at the same superficial velocity and returned to the feed chest. Air was first introduced through porous plates under the stirrer. Different bubble sizes between about 1 and 5 mm were obtained, at constant air flow by changing the surface of the porous air injection plates. Bubble size distribution was strongly affected by the stirrer velocity and could not be kept constant over a long period since the pores of the air injection plates tended to become wet or plugged Žbubble size increased as less pores were available for air injection.. Air was then introduced directly in the feed pipe through an air injection nozzle, which improved the reproducibility of the flotation results. Bubble size distribution remained however quite large, could not be modified easily and was strongly affected by the turbulence level at high stirrer speed. Bubble sizes were measured directly in the cell by a photographic method, under conventional flotation conditions in terms of liquid surface tension but without fibers Žpulp filtrate with chemicals.. Detailed characteristics of the first experimental cell and flotation results have been reported in previous publications ŽJulien Saint Amand, 1991; Julien Saint Amand and Perrin, 1991.. Typically the cell was operated at 0.5 mrs stirrer speed. The effects of pulp consistency, particle size and bubble size were investigated on repulped magazines with specks from UV varnished prints. Small bubbles between 1 and 2 mm in diameter proved to be much more efficient than large bubbles between 2 and 5 mm in diameter, since the brightness gain was doubled at about 300% air ratio. It was also shown that flotation under normal deinking conditions has first order kinetics with respect to the air ratio ŽFig. 1.. The second generation of experimental cell was designed to improve bubble size control at different turbulence levels. A variable speed stirring device was implemented

284

F. Julien Saint Amandr Int. J. Miner. Process. 56 (1999) 277–316

over the whole height of the cell as shown in Fig. 3, and a square section Ž150 = 150 mm2 . was chosen in order to avoid excessive vortex formation and to allow better flow visualization through the transparent front and rear walls. The cell was operated in recirculation to vary the cumulated air ratio. Pulp was introduced at the top of the cell and the accepts collected at the bottom and returned to the feed chest through a constant level device in order to adjust the suspension height over the inlet level. The stirring device was equipped with three cylindrical fins of 5 mm diameter located at different positions each 100 mm, as shown in Fig. 3. Under these conditions the experimental cell was operated in a downward pulp flow configuration at low superficial velocity Ž0.75 cmrs. in order to allow small bubbles to be removed, and the average turbulence was kept constant over the cell height. Air was first introduced at the bottom of the cell, such as in column flotation, in order to investigate the effect of turbulence at different bubble sizes and roughly constant contact time. However as bubble size proved to be strongly influenced by the flow velocity at the surface of the new air injection device, air was then mixed in the feed flow and thus introduced at the top of the cell, in such a way as to separately control bubble size and turbulence level. The drawback to this modification is that turbulence is exerted over a shorter and uncontrolled contact time as large bubbles are removed faster from the pulp suspension than small bubbles, which makes it difficult to quantitatively assess the effect of the turbulence generated mechanically in the cell. The new air injection device is based on several capillary tubes supplied with pressurized air and connected perpendicularly to the surface of a small injection plate. Under these conditions the air flow through each capillary tube depends only on the air feed pressure for constant tube diameter and length. By contrast in the case of porous plates, the air flow distribution is very unequal over the porous surface since the pores are different in diameter and length and also subject to a large influence of water penetration because of their small diameter and short length. As bubble size is known to depend on the capillary tube diameter, air flow and liquid surface tension, and has been shown to be strongly affected by the pulp flow along the plate surface, different air injection conditions were tested. Air injection plates with capillary tubes of different diameters and lengths were placed in a channel in such a way to study the influence of the shear flow along the plate surface. The results obtained with capillary tubes of 250- and 500-mm diameter and 1-m length in a channel of 5-mm thickness are shown in Fig. 4. Bubble sizes were measured directly in the channel on video pictures by using a flash to freeze the motion of the air bubbles. The measurements showed that bubbles of different sizes could be produced in a very narrow size range with the new air injection technique. The air injection technique was then adapted to the experimental flotation cell, in order to produce calibrated bubbles in the relevant size range of 0.5–2 mm in diameter. As a stable air supply could not be obtained under less than 0.3 b feed pressure, the flow velocity and shear had to be increased at the surface of the air injection plate in order to produce the small bubbles. One or two pulp feed channels of 2 or 5 mm thickness were supplied with 100 capillary tubes each in order to produce bubbles of different sizes as shown in Table 1.

F. Julien Saint Amandr Int. J. Miner. Process. 56 (1999) 277–316

285

Fig. 4. Bubble sizes vs. air feed pressure for different capillary diameters and flow velocities in the feed channel Žwater with surfactant, s s 50 mNrm..

Bubble sizes were measured in water with surfactant or in deinking process water which showed no significant difference. The presence of fibers showed no large effect on the bubble size according to some measurements, which were however not very accurate since the quality of the video pictures was reduced. As already mentioned, the video measurements were performed directly in the channel, about 5 cm after the bubbles had been formed and detached from the microtubes, which is assumed to lead to a slight underestimation of the real bubble sizes within the cell as the bubbles may further grow because of residual inertial forces at the measuring point. Bubble coalescence in the feed channel should not be significant since the distance is very short from the measuring point to the cell. It has also to be pointed out that the different flow velocities in the feed channel influence the ink collection conditions in the channel. The formation of bubbles at small orifices has been investigated experimentally as well as theoretically ŽIves, 1984.. At vanishingly small air flow rates the diameter Ž d b .

Table 1 Aeration conditions used for the flotation trials on the second experimental cell Channel thickness

2 mm

5 mm

Number of capillary tubes

100

200

200

100

200

200

Air feed pressure Žbar. One pass air ratio Ž%. Channel inlet velocity Žmrs. Channel outlet velocity Žmrs. Total air ratio Ž%.

0.3 3.8 4.2 4.3 0.5

0.3 9.7 2.1 2.3 0.9

0.6 36 2.1 2.8 1.2

0.3 3.8 1.7 1.7 0.9

0.3 9.7 0.8 0.9 1.2

0.6 36 0.8 1.1 1.8

286

F. Julien Saint Amandr Int. J. Miner. Process. 56 (1999) 277–316

of the bubble formed as it leaves the orifice Ždiameter d o . in a steady liquid is determined by a balance between the surface tension Ž s . and the pressure gradient due to gravity Ž g .:

p d o s s p d b3 D r gr6 or d b s Ž 6 d o srD r g .

1r3

Ž 8.

If surface tension can be neglected, the formation of the bubble is governed by the hydrodynamic forces. Theoretical analysis ŽDavidson and Schuler, 1960. considering the added and virtual mass of the bubble to be accelerated gave, after integration up to detachment, a simple result related to the air flow rate Ž Q . through the orifice: d b s 1.38 Ž Q 2rg .

0 .2

Ž 9.

The size of bubbles generated at the outlet of a 250-mm capillary tube is 1.96 mm according to Eq. Ž8. for a surface tension of 50 mNrm, and 0.92 and 1.56 mm, respectively, according to Eq. Ž9. for the air flow rates calculated from Table 1 at 0.3 and 0.6 b air feed pressure, respectively. The measured bubble diameters given in Table 1 agree quite well with the calculated values according to Eq. Ž9., as the bubbles are formed under dynamic conditions Ž0.4–1.5 mrs air velocity in the capillary tubes and 0.8–4.2 mrs in the channel.. The higher measured bubble diameters are assumed to be due to the surface tension Žneglected in Eq. Ž9.. while the lower values measured at high channel flow velocity are due to the shear forces.

4. Experimental results The flotation trials reported and discussed hereafter were performed on the latest experimental cell under usual deinking conditions with respect to pulping chemistry Ž1% soap, 1% caustic soda and 2.5% sodium silicate. and flotation consistency Ž1%. and temperature Ž408C.. Two test series were performed on different ink particles to investigate separately the influence of bubble size and turbulence. The results of the first test series performed on ink particles from magazines were analyzed in terms of global deinking flotation efficiency. The results of the second test series performed on different sizes of particles from the same inks had to be analyzed in terms of flotation rates, since bubble and particle size had a very large influence on flotation efficiency, and because flotation rates have to be considered in order to compare with flotation theory. 4.1. First test series (effect of bubble size) The aim of the first test series was to study the effect of bubble size and air ratio on deinking efficiency and pulp losses. Repulped magazines were used and floated under the usual deinking conditions. Turbulence in the cell was set at a low level Ž1 mrs peripheral stirring velocity.. The ink removal efficiencies were assessed globally for the whole ink particle size range by measuring the global ink content on pulp samples taken from the cell accepts at

F. Julien Saint Amandr Int. J. Miner. Process. 56 (1999) 277–316

287

Fig. 5. Removal efficiencies of detached inks from magazines vs. air ratio at different bubble sizes Žresults obtained with the 2 and 5 mm channel, respectively, represented by filled and open symbols..

different flotation times. Measurements of the Effective Residual Ink Concentration were performed on filter pads by using the ERIC device based on differences in light absorption and diffusion of inks and fibers in the visible and infra-red spectra ŽJordan and Popson, 1964.. The flotation efficiencies were calculated for the detached inks. Therefore the ERIC measurements were performed on the fiber fraction Žobtained by washing out the detached inks on a 100 mm screen. to assess the amount of attached inks, as well as on the whole pulp, in order to calculate the amount detached inks by difference. The effect of bubble size and air ratio on ink flotation efficiency is shown in Fig. 5. The curves drawn for the four different bubble sizes correspond to the six air injection conditions given in Table 2. They correspond for each trial, to three flotation times Žcalculated with respect to the total pulp volume remaining in the feed chest and flotation cell, for 1 pass air ratio q50, 100 and 200% air fed at 0.3 b and for 1 pass air ratio q100, 200 and 400% air fed at 0.6 b.. When two different aeration conditions were used for a same bubble size Ž0.9 and 1.2 mm produced with different channel thickness and velocity. the results showed no large differences.

Table 2 Surface properties of the laser inks used for the flotation trials Measuring method

Owens–Wendt

Surface energy ŽmJrm2 .

gsp

gsd

gs

g Sy

Acid–base

g Sq

gS

Toner A Toner B

0.1 2.8

22.4 26.3

22.5 29.1

0.8 0.1

1.1 6.4

24.9 36.0

288

F. Julien Saint Amandr Int. J. Miner. Process. 56 (1999) 277–316

The relations between ink flotation efficiency and bubble size in Fig. 6 are drawn from Fig. 5 for 100 and 200% air ratio. The results show the large influence of bubble size. Flotation efficiency increased from about 30% to 80% as the bubble size was decreased from 1.8 to 0.5 mm, for about 150% air. The higher efficiencies obtained with small bubbles are mainly due to the large increase in bubble number at a given air ratio, though less particles are collected by each bubble. For the first test series the flotation rejects were removed by foam overflow. This led to quite high pulp losses since only a little foam was formed because of the low superficial air flow velocity Ž0.3–3 mmrs. and had to be removed from the cell with the aid of air blown at the surface. Lower reject rates were obtained in the second test series, without foam overflow, by using a small suction tube for discontinuous removal of the rejects as foam had accumulated at the surface. The results in Fig. 7 show the variations of pulp losses with respect to bubble size and air ratio, under the same flotation conditions. The amount of flotation rejects increases with flotation time and as bubble size is decreased. The large increase in pulp losses observed as bubble size is decreased from 0.9 to 0.5 mm was due to higher fiber losses. Focusing on the fillers and fines rejected in the foam Žfraction of rejects passing a 100-mm screen. with respect to the amount of fillers and fines in the initial pulp, the flotation losses were much higher as shown in Fig. 8. Compared to the variations of pulp losses, the variations of fillers and fines losses were similar with respect to bubble size and air ratio, except for the 0.5-mm bubbles showing a lower increase since fiber losses were excluded. In general the variations in fillers and fines losses with respect to bubble size and air ratio were very similar to those observed for ink removal. The flotation results obtained in terms of detached ink removal efficiency and fillers and fines losses are summarized in Fig. 9. For each flotation test performed with

Fig. 6. Removal efficiencies from Fig. 5 vs. bubble size at 100 and 200% air ratio.

F. Julien Saint Amandr Int. J. Miner. Process. 56 (1999) 277–316

289

Fig. 7. Pulp losses corresponding to Fig. 5 vs. air ratio at different bubble sizes Žresults obtained with the 2 and 5 mm channel, respectively, represented by filled and open symbols..

different bubble sizes between 0.5 and 1.8 mm, the flotation efficiencies were reported vs. fillers and fines losses for 100% and 200% air ratio. The results clearly show a direct correlation between removal efficiencies of detached ink particles and fillers and fines, as far as the foam removal conditions are kept constant. Such results suggest that particle size distribution may have a larger influence on the flotation kinetics than certain differences in particle surface properties, with respect to

Fig. 8. Fillers and fines losses corresponding to Fig. 7 vs. air ratio at different bubble sizes Žresults obtained with the 2 and 5 mm channel, respectively, represented by filled and open symbols..

290

F. Julien Saint Amandr Int. J. Miner. Process. 56 (1999) 277–316

Fig. 9. Removal efficiency of detached inks from magazines vs. fillers and fines losses for different bubble sizes and for 100 and 200% air ratio.

bubble size, since the ink particles and fillers and fines are in the same size range. However, as already mentioned, the reject were removed manually by foam overflow, with a small foam height so that some of the liquid suspension from the surface was also rejected. Under these conditions the fillers and fines losses are overestimated since they are assumed to include particles collected by the air bubbles as well as material brought to the surface in the boundary layer and wake of the bubbles but not really collected onto the bubbles. Lower reject rates were obtained in the second test series, since the foam removal conditions were improved by using a suction device in order to avoid losses from suspension at the surface. 4.2. Second test series (effect of turbulence) The aim of the second test series was to investigate the effect of turbulence on flotation kinetics in relation to bubble size and particle size. Unprinted magazine papers were repulped with laser inks Žcalibrated particles added in the pulper a few minutes before the end of pulping. and floated under the same normal deinking conditions Ž1% soap, 1% caustic soda, 2.5% sodium silicate, 1% consistency and 408C.. The different sizes of calibrated laser inks were produced by melting the toner, breaking up the molten ink and screening the particles on different wire screens with openings between 75 and 500 mm. Particle sizes in the microscopic and visible size range were used: . Microscopic toner particles mainly between about 5 and 20 mm. . Small visible particles calibrated between 75 and 100 mm. . Large visible particles calibrated between 300 and 500 mm.

F. Julien Saint Amandr Int. J. Miner. Process. 56 (1999) 277–316

291

Two different laser inks were tested under the same conditions. The particle size distribution of the two toner inks were very similar as shown by the image analysis measurements in Fig. 10. The mean size of the toner particles was about 10 mm. The two laser inks also had similar densities, of about 1.2 for toner A and close to 1 for toner B. The surface properties of the two laser inks were measured on deposits obtained by melting the toner particles, according to the Owens–Wendt and acid–base methods. Both laser inks showed quite low surface energy and hydrophobic character as shown in Table 2. Despite some differences of surface energy and components, similar flotation kinetics were obtained on the two laser inks of the same particle size range. This similar behavior is believed to be due to the fact that the surfactant Žsoap. adsorbed on the particle surface neutralizes the effect of differences in particle surface properties with respect to the collection mechanism of particles onto the bubbles. As a consequence of the similar behavior of the two laser inks and in order to improve the accuracy of the results, all the rate constants reported hereafter are mean values of two flotation tests performed on each type of laser ink under the same flotation conditions. Samples were taken from the accepts at different flotation times according to given air ratios which were adapted to the flotation conditions Žbubble size and particle floatability. in order to facilitate the assessment of the rate constants. A decrease of flotation rates was generally observed beyond about 90% particle removal even in the case of calibrated particles. This was assumed to be due to a low proportion of less floatable particles having for example different shapes. No significant increase of the surface tension Ž56–58 mNrm., as might be expected when surfactants are removed in the foam, was observed during flotation. A method has been proposed to calculate the flotation rate constant according to modified first order kinetics taking into account the fraction of unfloatable particles

Fig. 10. Particle size distribution of the toner inks used for the flotation trials.

292

F. Julien Saint Amandr Int. J. Miner. Process. 56 (1999) 277–316

ŽDorris and Page, 1995.. Since the particle concentration measurements were generally not accurate for statistical reasons Žfew particles counted. at the end of the flotation, mean values of the last points were used to calculate the rate constants. The flotation tests on the different types of inks and particle sizes were performed at two turbulence levels and with two bubbles sizes of 0.5 and 1.8 mm. The two ‘turbulence levels’ of 1 and 2 mrs are represented by the stirring velocity calculated at the end of the radial fins. A lower stirring velocity than about 1 mrs left some dead zones in the flotation cell, while a velocity higher than about 2 mrs produced to much internal recirculation. The flotation results illustrated in Figs. 11 and 12 show a large influence of particle size on the flotation kinetics of laser inks as well as on the effect of turbulence. The flotation rate constants of the microscopic particles were much lower than those of the particles in the visible size range. The largest particles also showed high flotation rates. This is not in agreement with most of the results published about the effect of particle size. Industrial specks of different size, shape, density and surface properties are to be found in deinking pulps and it is generally considered that, on average, large specks are the most difficult to remove by flotation ŽMac Cool and Leblanc, 1989.. In the specific case of laser prints, the poor floatability of the larger particles has been shown to be due to fibers retained in the toner particle which increase the ‘hydrodynamic size’ of such ‘hairy’ particles ŽPan et al., 1994; Dorris and Page, 1995; Vidotti et al., 1995.. Nevertheless the results indicated that even though the presence of fibers in toner specks retarded flotation, removal rates of these particles remained high, which is assumed to be due to their naturally hydrophobic surface properties.

Fig. 11. Flotation rate constants of laser inks obtained with large bubbles of 1.8 mm diameter vs. particle size range and turbulence level Ž1 and 2 mrs stirring speed..

F. Julien Saint Amandr Int. J. Miner. Process. 56 (1999) 277–316

293

Fig. 12. Flotation rate constants of laser inks obtained with small bubbles of 0.5 mm diameter vs. particle size range and turbulence level Ž1 and 2 mrs stirring speed..

In the case of flat shaped particles, the strong decrease of the flotation rate constant observed for the large specks on our first experimental cell ŽJulien Saint Amand and Perrin, 1991. corresponded to less hydrophobic particles Žabout 45 mJrm2 surface energy according to the Owens–Wendt method.. Higher flotation rates were obtained later on with more hydrophobic specks from different UV varnished prints. In any case the flat shape of the particles is considered to be responsible for lower flotation rates ŽSchmidt and Berg, 1996.. Very low flotation rate constants were obtained with the large bubbles on the microscopic toner particles Ž5–20 mm. as shown in Fig. 11. At low turbulence the rate constants of the small particles Ž75–100 mm. were about 10 times higher than for the microscopic particles and remained high for the large particles Ž300–500 mm.. Concerning the effect of turbulence the results showed a strong dependence on particle size. A turbulence increase had a large positive effect on the microscopic particles, the flotation rate constant being about two times higher. On the particles in the visible size range, the turbulence increase had a less positive effect on the flotation kinetics. These differences with respect to particle size may be due to the fact that particle bubble collision is promoted by turbulence, whereas adhesion and stabilization onto the bubble are affected inversely. The results in Fig. 12 obtained with the small 0.5-mm bubbles showed roughly the same variations with respect to particle size as those in Fig. 11 obtained with large bubbles. However much higher values of the flotation rate constants were observed, which confirms that small bubbles are more efficient than large bubbles at a given air ratio. A turbulence increase showed also a large positive effect on the microscopic particles, almost no effect on the small particles and a negative effect on the large particles.

294

F. Julien Saint Amandr Int. J. Miner. Process. 56 (1999) 277–316

As already discussed for the large bubbles, these results may reflect the fact that higher collision number achieved by increased turbulence does not compensate for reduced collection probability, especially in the case of small bubbles, since they are dragged more completely by the turbulent eddies than large bubbles. Consequently the higher mean acceleration increases the relative bubble velocity and thus reduces the contact time which may become to short to achieve drainage and rupture of the thin liquid film between the particle and the air bubble. In addition, large particles are more subject to detachment than small particles in turbulent flow, especially in the case of small bubbles as illustrated in Fig. 15. The flotation rejects were checked during the flotation trials. Pulp losses and ash content of the rejects were shown to be reproducible over each series of six tests Žthree particle sizes and two types of toner inks. under given flotation conditions Žbubble size and turbulence level. despite the manual foam removal technique Ždiscontinuous foam suction.. The mean values of pulp and filler losses are given in Fig. 13 for 100% air ratio. On average the flotation losses were three and four times lower than reported in Figs. 7 and 8, because of the improved foam removal technique. The variations with respect to bubble size were similar. Higher losses were observed under the most efficient flotation conditions, with the small bubbles. Filler losses were about three times higher than pulp losses. Concerning the effect of turbulence, a positive influence was observed with the small bubbles, especially on the pulp losses with respect to filler losses, which indicates that fiber losses were reduced at increased turbulence. Assuming that fibers are entrained into the foam by small bubbles weakly attached on fibers or entrapped in flocs, a turbulence increase should reduce this fiber loss mechanism and thus explain the

Fig. 13. Pulp and filler losses on repulped magazine papers after flotation at 100% air ratio vs. bubble size and turbulence level Ž1 and 2 mrs stirring speed..

F. Julien Saint Amandr Int. J. Miner. Process. 56 (1999) 277–316

295

experimental results. Moreover, hydraulic entrainment, a mechanism extensively investigated in mineral flotation which has been shown to be the dominant fiber loss mechanism ŽAjersch and Pelton, 1995., should also be reduced by a turbulence increase as the wake of the bubble may be destabilized by the turbulent eddies. With large bubbles, the turbulence showed the opposite effect on pulp and filler losses. This conflicting observation suggests that other phenomena have to be considered. In fact it has been observed that with large bubbles the higher turbulence level produces some bubble fragmentation, which seemed not to be the case with the small bubbles. Under these conditions the additional small bubbles produced by the increased turbulence may explain the increase in flotation losses observed with the large bubbles.

5. Theoretical analysis The results of the second test series Žexpressed in terms of flotation rates and related to well defined particles and bubbles. were analyzed with respect to the predictions of the theory regarding the influence of bubble size, particle size and turbulence. This theoretical analysis required to evaluate the flow characteristics in the aeration channel and in the cell body since the particle bubble collision and attachment probabilities are strongly related to the turbulence. 5.1. Cell flow characteristics According to the aeration conditions in Table 1, and to the channel thickness Ž y . and length Ž l s 100 mm. after air injection, the flow in the channel is just above the turbulent limit Ž5 = 10 3 - Re s Uyrn - 10 4 ., the mean shear rate Ž G s 2Ury . in the range of 3 = 10 2 to 4 = 10 3 sy1 and particle bubble contact time within the channel between 20 and 100 ms. In laminar shear flow the energy dissipation is given by ŽBird et al., 1960.:

´ s n G2

Ž 10 .

which is equal to 0.1 to 20 Wrkg in the channel. In isotropic turbulent flow the energy dissipation approximates to ŽHinze, 1959.:

´ f uX 3rL

Ž 11 .

where uX is the RMS turbulent velocity and L the spatial integral scale related to the large eddies. Eq. Ž11. gives an energy dissipation of the same order as Eq. Ž10. for about 7% turbulence and a value of L f yr5 which is in agreement with boundary layer thickness evaluations ŽPlate and Schulze, 1991.. At the outlet of the aeration channels one can assume that most of the kinetic energy Ž r U 2r2. of the pulp and air jet is lost in the cell and calculate the dissipated power Ž QD P . from the pressure drop and pulp flow rate. The power dissipation varies between 0.05 and 1.5 W for the conditions given in Table 2. A relation has been proposed

F. Julien Saint Amandr Int. J. Miner. Process. 56 (1999) 277–316

296

ŽSchulze, 1994. to evaluate the maximum energy dissipation in a turbulent free jet Ždiameter d j .:

´ s 0.13U 3rd j

Ž 12 .

which gives 9 to 2700 Wrkg for the jet at the outlet of the channel by taking the hydraulic diameters Ž3.6 and 8 mm.. The same range of energy dissipation is obtained by using Eq. Ž11. for L f yr5 and about 25% turbulence in the mixing layer at the outlet of the channel, as typically observed for plane jets ŽBinder, 1993.. Such a high value of the upper limit of the energy dissipation range corresponds to the assumption that all the kinetic energy is dissipated at the outlet of the channel in a volume of same section over a length of about seven times the channel thickness. In fact the turbulence dissipation volume is higher because of the jet expansion, and the mean energy dissipation consequently much lower than the local value calculated from Eq. Ž12.. These values have to be compared with the power dissipated in the cell by the stirrer. Assuming that the flow motion induced by the rotation of the radial fins of the stirrer Žradial flow patterns and reduced flow rotation because of the square section of the cell. can be neglected, the stirring power can be calculated from the drag forces integrated over the length of the radial fins Ž120-mm stirring wheel comprising three cylinders of 5-mm diameter.: 2

P s v rpCxr2 Ž v r . ddr f r Cxr8 v 3 dr 4

H

Ž 13 .

the drag coefficient of a cylinder being roughly constant ŽComolet and Bonnin, 1961. for the Reynolds numbers calculated over the peripheral portion of the radial fins in the experimental stirring velocity range Ž v rdrn - 10 4 for v r - 2 mrs, Cx f 1 for 10 3 Re - 10 4 .. This gives a power dissipation of about 1 W for each stirring wheel at 2 mrs stirring velocity. According to these evaluations, the total power dissipated to promote particle bubble contact in the cell is about 3 W at 2 mrs stirring velocity and 0.75 W at 1 mrs for the three upper stirring wheels involved in particle bubble collision Žsee Fig. 1., while about 1.5 W is dissipated at the outlet of the channel under the conditions used to produce the small 0.5-mm bubbles. Assuming that most of the turbulence dissipated around the cylindrical fins and in the wakes is located in the volume swept out by the rotation of the fins, the mean energy dissipation in this volume varies between 5 and 20 Wrkg for 1 s residence time. If one considers the whole volume where particle–bubble collision may occur, the contact time is increased to about 30 s and the average energy dissipation reduced in the same proportion. Other relevant characteristics of the turbulence generated by the stirrer are the scale, velocity fluctuations and characteristic time related to the large eddies. From the fluid lines around a cylinder ŽComolet and Bonnin, 1961. at Reynolds numbers between 10 3 and 10 4 it can be shown that the maximum acceleration of the fluid around the fins is in the order of g s U 2rd Ža velocity variation U over a time drU ., which gives 200–800 mrs 2 for 1–2 mrs stirring speed.

F. Julien Saint Amandr Int. J. Miner. Process. 56 (1999) 277–316

297

From Batchelor, as cited in the work of Abrahamson Ž1975., the mean acceleration Žg . in a turbulent flow field is related to the energy dissipation Ž ´ . by:

g 2 s 1.3 Ž ´ 3rn .

1r2

Ž 14 . 2.

which gives accelerations of the same order Ž130–370 mrs for 5–20 Wrkg calculated from the drag force on the cylindrical fins, and therefore including the smaller eddies in the wake. The scale of these eddies should be less than about 1 mm from experimental data about plane wakes ŽBinder, 1993.. For a rough evaluation of the characteristic time of strong large scale eddies generated by the stirrer to promote particle bubble collision, one can consider the flow around the fins and take drU, which gives less than 5 ms for U ) 1 mrs. In turbulent flow fields the characteristic time for velocity fluctuations over a distance l is given by ŽAbrahamson, 1975.:

t s 0.75´y1 r3 l 2r3 f 0.75uX 2r´ , for l s L

Ž 15 .

which gives 13 and 8 ms, respectivel, for l s d and ´ s 5–Wrkg calculated from the volume swept by the fins Žat U s 1 and 2 mrs.. Considering the mean characteristics of the turbulence in this volume and assuming that the scale of the large eddies Ž L f l f d s 5 mm. remains roughly constant, the turbulence characteristic time, given by the relation t f LrUm from Taylor’s hypothesis ŽFortier, 1967., is in the same range for a mean fluid velocity Um f Ur3. From Eq. Ž11., with L f d the turbulent velocity uX is then about 0.3–0.45 mrs and the mean acceleration is in the range of 20–60 mrs 2 since g f uXrt in a turbulent flow ŽHinze, 1959., which is in agreement with Eqs. Ž11. and Ž14. by using the constant 0.75 in the equation of energy dissipation Ž ´ s 0.75 uX 3rL. according to Hinze ŽHinze, 1959; Weiss and Schubert, 1987.. Some quite large variations are found in the assessment of the turbulent flow characteristics around the stirrer according to the above calculations, as the assumption of steady flow with respect to the fluid is not valid and because the turbulence scale taken for the calculations Ž L f l f d s 5 mm. is only an hypothesis and not constant in the considered volume. It also must be emphasized that most of the theory of turbulence refers to orders of magnitude. Under these conditions and with the assumption that the velocity of the fluid around the stirrer wheels is about one third of the velocity of the fins ŽUm s Ur3., a better evaluation of the turbulence characteristics for the same hypothesis on the turbulence scale Ž L f l f d . gives roughly Žthe maximum acceleration and the smallest eddies being located around the fins, including the wake.: . Turbulence scale: about 5 mm and less, . Energy dissipation: 2–10 Wrkg, . Maximum acceleration: 100–400 mrs 2 , . Mean acceleration: 20–100 mrs 2 , . Turbulent RMS velocity: 0.2–0.5 mrs, . Characteristic time: 5–20 ms. Between the stirring wheels the particle–bubble contact time is much longer Žabout 5 min instead of 1 s., the energy dissipation much lower Žin the order of 10y2 Wrkg., the

F. Julien Saint Amandr Int. J. Miner. Process. 56 (1999) 277–316

298

mean turbulent acceleration becomes significantly lower than gravity acceleration, and the characteristic time may be in the order of a few tenths of a second for a turbulence scale in the range of 1 cm and for the same velocity fluctuations of about 0.2–0.5 mrs. 5.2. Effect of flotation parameters The effects on flotation rates of hydrodynamic parameters such as turbulence and number and size of bubbles and particles, are given by Eq. Ž1. ŽAhmed and Jameson, 1989; Schulze, 1994., where the individual probabilities Ž Pc , Pa and Ps . of the subsequent collision, attachment and stabilization processes have been discussed in the complete review of flotation theory given by Schulze Ž1994. ŽEq. Ž1... This simplified kinetic equation does not take the balance between free and attached particles into account, since the probability Ž1 y Ps . that a particlerbubble aggregate becomes unstable applies on the number of these aggregates while the cumulated probability Ž Pc Pa Ps . applies on the numbers of free particles and bubbles ŽPlate and Schulze, 1991; Heindel, 1997.. It is suggested that particlerbubble detachment should depend on the frequency of external stress in the turbulent flow field which should be related to the collision frequency. The theoretical analysis of the experimental results was however limited to the comparison with the predictions based on Eq. Ž1. and its modified form ŽEq. Ž4.. proposed previously, which should be valid under the tested conditions. The flotation rate constant Ž K . determined experimentally is related to the theoretical probabilities by Eq. Ž5. calculated from Eq. Ž4. which is based on contact time Ž t c . and air ratio ŽT . instead of flotation time Ž t . and bubble number Ž Nb .. Under turbulent flow conditions where inertial effects determine the relative velocity of particles about to collide, the number of particle–bubble collisions per unit time and volume is given by the following expression ŽSchulze, 1994; Ahmed and Jameson, 1989. adapted from Abrahamson’s analysis ŽAbrahamson, 1975.: Z s Ž 8p .

1r2

2 Np Nb d pb Ž Õtp2 q Õtb2 .

1r2

Ž 16 .

where d pb s Ž d p q d b .r2 is the contact distance and Ž Õti2 .1r2 s 0.33 ´ 4r9 d i7r9ny1r3 Ž D rrr . 2r3 is the effective RMS value of relative velocity between particle or bubble, and the fluid. As the bubbles are generally much larger than the particles to be collected Ž d b 4 d p . and have a higher density difference Ž D rrr . with respect to the fluid compared to ink particles, d p and Õtp2 can be neglected in Eq. Ž16. which becomes: Z f Ž pr2 .

1r2

Np Nb d b2 Ž 0.33 ´ 4r9 d b7r9ny1r3ry2r3 . f 0.41 Np Nb d b25r9´ 4r9ny1r3

Ž 17 . For smaller particles and bubbles, i.e., if d pb is small compared with the smallest eddies in the fluid and particles follow the fluid motion completely, collision is controlled by the velocity gradient within the eddies and particle inertial forces are neglected. Calculations based on the Smoluchowski mechanism for gradient collision in laminar

F. Julien Saint Amandr Int. J. Miner. Process. 56 (1999) 277–316

299

flow, the gradient being expressed by Eq. Ž10. from Kolmogoroff’s dimensional analysis, gave ŽAbrahamson, 1975.: Z s Ž 8pr15.

1r2

3 Np Nb d pb Ž ´rn .

1r2

Ž 18 .

As the collision rates have been evaluated in the turbulent flow, one has to consider the collision probability which is governed by the flow lines around the bubble assuming that the particle is much smaller. The equation established for the collision probability thus includes the bubble Reynolds number ŽRe b . to describe this flow field: Pc s Ž 3r2 q 4 Re 0b .72r15 . d p2rd b2

Ž 19 .

This equation is valid for particles smaller than 100 mm and bubbles smaller than 1 mm having rigid interface because of adsorbed surfactants ŽYoon, 1991; Schulze, 1994.. The probability of attachment in flotation deinking is mainly described by the sliding process of the particle along the bubble surface since the ink particle density is relatively low with respect to the density of mineral particles where the collision process is more relevant ŽSchulze, 1994; Nguyen-Van et al., 1995.. The probability has been evaluated from calculations based on the film drainage time necessary to reach the critical thickness, with respect to the sliding time. As the equations were solved numerically ŽSchulze, 1992. no relation is available for the attachment probability Pa as well as for the stabilization probability Ps . Nevertheless, it has been reported ŽSchulze, 1994. that for given physico-chemical conditions Žparticle surface properties, flotation chemistry and temperature. the attachment probability is reduced as the sliding time decreases, i.e., for large bubbles and higher acceleration, and also as particle size is increased because of the larger film area to be drained off. Concerning the stabilization probability the reported results ŽSchulze, 1994. showed that the probability of detachment can be neglected for low density particles Ž1.1 grcm3 . and size Ž- 200 mm., even at quite high energy dissipation Ž130 Wrkg. and for very small bubbles Ž0.3 mm. with respect to the relevant flotation deinking conditions.

6. Discussion The investigations carried out on the experimental cell to assess the influence of bubble size, particle size and turbulence on the flotation kinetics refer to calibrated bubbles between 0.5 and 1.8 mm, rather spherical laser ink particles between 5 and 500 mm, and variable turbulence in time and space as the pulp suspension moves from the inlet channel to the separation zone. The turbulence characteristics were evaluated in the cell in order to discuss the effects in relation to the flotation theory. Considering the characteristic time of particles in turbulent flow, which is of the order of r b2rn for bubbles Ž r b s d br2. according to Fortier Ž1967. and gives 0.08 to 1 s for bubbles between 0.5 and 1.8 mm, we conclude that the bubbles are not entrained by the high turbulence created around the fins and that only the small 0.5-mm bubbles reach their terminal velocity with respect to the fluctuations of turbulent acceleration.

300

F. Julien Saint Amandr Int. J. Miner. Process. 56 (1999) 277–316

Consequently, only gravity collision to be observed mainly in the volume between the stirring wheels should be discussed in relation to the equations giving the probability of collision. The experimental results will thus be difficult to analyze since turbulent collision generated around the fins is believed to contribute significantly to particle– bubble collision. The flotation rate constants obtained at low stirring speed for small and large ink particles are shown in Fig. 14 in relation to the bubble diameter. From the theoretical relations, giving the collision rate for inertial approach mechanisms ŽEq. Ž17.. and for gradient collision ŽEq. Ž18.. and combined with Eqs. Ž1. and Ž5., it follows respectively: K s 12 t c Ž 2rp .

1r2

2 y3 d pb d b Ž Õtp2 q Õtb2 .

1r2

Pc Pa Ps f 0.8t c dy2r9 ´ 4r9ny1r3 Pc Pa Ps b

Ž 20 . K s 12 t c Ž 2r15p .

1r2

3 Ž d pbrd b . Ž ´rn . 1r2 Pc Pa Ps f 0.3tc Ž ´rn . 1r2 Pc Pa Ps Ž 21.

which gives the following relation Žfor d b 4 d p .: K A d bn ´ r Pc Pa Ps

with y0.22 - n - 0 and 0.44 - r - 0.5

Ž 22 .

Assuming that particles and bubbles are small compared to the smallest eddies, which should be the case in the cell volume between the stirring fins where gravity collision is observed, the bubble velocity to be used for the calculation of the Reynolds number in Eq. Ž19. is roughly given by the terminal velocity determined in water with surfactants ŽIves, 1984.. It follows from these rising velocities, i.e., 5–15 cmrs for 0.5–1.8 mm .9 bubbles, that the collision probability Pc is roughly proportional to dy0 and to d p2 . In b

Fig. 14. Flotation rate constants of small and large laser inks, magazine inks and varnish specks obtained at 1 mrs stirring speed vs. bubble size.

F. Julien Saint Amandr Int. J. Miner. Process. 56 (1999) 277–316

301

.46 addition it has been shown ŽYoon, 1991. that Pc varies as dy0 for very large bubbles b y2 and as d b for very small bubbles. Under the flotation conditions in the experimental cell, and according to a detailed analysis of available theory, the influence of bubble diameter d b on the probability of collision Pc and on the probability of adhesion by sliding Pa depends on the particle diameter d p ŽSchulze, 1996.:

Pc A dy1 b

for

25 - d p - 50 mm

Pc A dy0.75 b

for

50 - d p - 200 mm

Pc A dy0.35 b

for

d p ) 200 mm

Pa A dy0.75 b

for

d p - 50 mm

Pa A dy0.63 b

for

d p ) 50 mm

the relations about adhesion probability being only valid for bubbles with a rigid surface as observed with surfactants Žthe dependence of Pa on bubble size is higher in the case of mobile bubble surface.. Assuming that the probability of detachment is low Ž Ps f 1. which should be the case for the small particles ŽSchulze, 1994., the relations between flotation rate constant and bubble size are then: K A d bn y1.97 - n - y1.75 y1.20 - n - y0.98

with: for small particles for large particles

The higher limits of these n values are in quite good agreement with the experimental results shown in Fig. 14, which means that the mean collision rate under the flotation conditions of the test may be better described by gradient collision ŽEq. Ž18.., andror that the negative effect of bubble size on the collision probability may be overestimated in theoretical calculations based on hypothesis such as those leading to Eq. Ž19.. This latter assumption is believed to be quite realistic if one considers the characteristic time of the bubbles in the experimental size range with respect to the characteristic time of the large scale turbulence, showing that bubbles do not reach their terminal velocity. Under these conditions the velocities of the bubbles with respect to the surrounding fluid submitted to turbulent and gravity acceleration are lower than assumed in the theoretical calculations. The sliding time is thus increased with respect to the film drainage time and the collision probability less affected by the bubble size increase. Experimental results ŽAhmed and Jameson, 1989. reporting flotation rates of latex particles with different bubble sizes at medium turbulence, gave n values between y1.1 and y1.5 for particles between 4 and 30 mm and bubbles between 75 and 650 mm. At high turbulence, lower n values between y0.8 and y1 were reported. These results are comparable to our results obtained on laser particles in the same size range Ž n s y1.5 at low stirring speed., and concerning the effect of turbulence Žlower influence of bubble size at high stirring speed.. On average, in a large particle size range and from all the experimental results obtained in the first and second test series, it is concluded that flotation rate constants

302

F. Julien Saint Amandr Int. J. Miner. Process. 56 (1999) 277–316

are roughly proportional to the inverse of the bubble diameter Ž n s y1. which means that the specific phase interface Žtotal bubble area per unit pulp volume. is, as recently suggested ŽGottsching et al., 1995., a most relevant parameter to characterize the effect ¨ of bubble size and air ratio. Concerning the effect of particle size on the flotation rate Ž K A d pm ., Eq. Ž19. and the calculations given in the work of Schulze Ž1994. for a low particle density Ž rp s 1.1 grcm3 . as in the case of laser inks, suggest that the flotation rate constant is proportional to d p Ž m s 1., while n values of about 0.7 have been reported for flotation trials on latex particles with small bubbles of about 0.5 mm under agitation ŽAhmed and Jameson, 1989.. Values of less than unity Ž0.45–0.65. were reported for toner particles in the microscopic size range, indicating a low dependence of the rate constant on particle size ŽDorris and Page, 1995.. The experimental results shown in Figs. 11 and 12 indicate a similar influence of particle size in the range of 10–100 mm. The values of m are between about 0.5 and 0.7 for the small 0.5-mm bubbles and between about 0.8 and 1.1 for the large 1.8-mm bubbles, respectively, the lower values of m being obtained at low stirring speed. In addition it has to be pointed out that the large particles used in the experiments were broken particles with a rather spherical shape and a certain surface roughness. This makes it difficult to compare experimental and theoretical results with respect to the influence of particle size, since it has been shown that particles with sharp edges are more easily captured by bubbles than spherical particles by accelerating the rupture of the disjoining liquid film ŽNguyen-Van et al., 1995.. For particles larger than about 100 mm, no further increase of the flotation rate is observed with particle size. This observation, commonly reported in the field of flotation deinking where large and usually flat shaped specks show a large efficiency drop with particle size, is believed to be mainly due to the detrimental effect of turbulence on the stabilization of the particle bubble aggregate. Fig. 15 showing particles and bubbles of sizes used for the trials and fluid lines drawn according to the bubble Reynolds numbers illustrates possible mechanisms of particle bubble collection and detachment.

Fig. 15. Schematic fluid lines around particles and bubbles drawn at scale in a turbulent flow field.

F. Julien Saint Amandr Int. J. Miner. Process. 56 (1999) 277–316

303

In the case of small 100-mm particles caught at the front side of a large 1.8-mm bubble and entrained along its surface, attachment is achieved if the sliding time is larger than the film drainage time. Once collected onto the bubble, small particles with respect to bubble size are swept to the back of the bubble, where they are protected from further detachment because of the back flow vortices in the wake. By contrast, in the case of large particles with respect to the bubble size, the relative contact time under turbulent accelerations may be too low to achieve particle–bubble attachment and stabilization. Moreover the particles collected onto the bubble are more subject to detachment under external stress produced by the turbulent flow field since large particles are not caught in the fluid lines around the bubble. For very small ink particles in the range of 1 mm, previous results have shown that the efficiency decrease is much lower than predicted by the theory or by extrapolation of the experimental results obtained on laser ink particles between 10 and 100 mm, which is believed to be due to the fact that very small ink particles may be flocculated and thus removed as larger particles ŽJulien Saint Amand and Perrin, 1991.. These ink agglomeration mechanisms depend mainly on deinking chemistry but also on turbulence level and particle size, since large flocs including large particles should not be produced at high turbulence. Physico-chemical interactions are thus believed to be of critical importance as far as optimization of turbulence with respect to particle size is concerned. Concerning the effect of turbulence on the flotation rate of the small particles of 10 mm, the results in Figs. 11 and 12 show a rate constant increase which is roughly proportional to the stirring speed. This is in agreement with the influence of the energy dissipation on the collision rate given in Eq. Ž18. Ž Z A ´ 1r2 A U .. However, as already discussed, the turbulence shows strong variations in time and space as the pulp suspension enters the aeration channel and cell body, which makes it very difficult to assess the contribution of these different turbulence states on the collision rate and thus to analyze the results. In addition it has to be kept in mind that the experimental results are not accurate enough to conclude for instance that gradient collision is more relevant that inertial collision. The effect of pulp consistency was not investigated on the second experimental flotation cell. Previous studies on the first experimental flotation cell have shown that the effect of consistency on the flotation rate constant increased as particle size and consistency were increased ŽJulien Saint Amand and Perrin, 1991.. The flotation rate of microscopic ink particles was not significantly affected in the tested consistency range Ž5–20 grl.. The influence of the fibers on the terminal rise velocity of the bubbles should not be very large under the experimental conditions discussed, since the fiber consistency was low Žabout 6 grl.. Further research is however necessary to investigate the interactions between fibers, ink particles and bubbles in a turbulent flow field. 7. Conclusion Experimental studies on hydrodynamics of flotation deinking have been carried out on laboratory flotation cells specially designed to investigate separately the influence of bubble size and turbulence on flotation efficiencies. Calibrated bubbles of different sizes

304

F. Julien Saint Amandr Int. J. Miner. Process. 56 (1999) 277–316

were produced by using capillary tubes for the air supply and calibrated laser ink particles without attached fibers were used for the flotation trials in order to assess the effects of particle and bubble sizes. Concerning the aeration conditions, it was shown that high flotation efficiencies are achieved at high air ratios according to first order kinetics, and with small bubbles as far as they can be removed from the cell. On average for the flotation of particles in a large size range, the specific phase interface between air and pulp was considered as one of the most relevant parameter which should be increase to optimize the aeration conditions. In general, the best aeration conditions in terms of ink removal efficiency also led to the highest flotation losses. However, no specific investigations were devoted to the optimization of pulp losses with respect to bubble size distribution, since the laboratory flotation conditions were not adapted to such studies. In terms of particle size, the efficiencies were much lower for the microscopic ink particles than for particles in the size range of the visible limit, in the specific case of detached laser inks having low surface energy and a rather spherical shape. No large efficiency drop was observed on the largest laser ink particles contrary to the general mill experience relating to specks, because of the favorable shape and surface properties of the particles tested. With respect to particle size, the influence of bubble size was shown to be higher for the smallest particles, which means that large bubbles are relatively more convenient for floating the largest particles in terms of flotation costs, since small bubbles require more energy and are more difficult to remove from the fiber suspension. However, small bubbles remain much more efficient at a given air ratio because of the higher specific phase interface. Concerning the effect of turbulence, the flotation results showed that high turbulence is necessary to remove the microscopic ink particles efficiently, while no positive effect of turbulence was observed on the large particles especially with small bubbles. It was also observed that fiber losses are reduced as turbulence is increased. It is thus believed that small individual ink particles require high turbulence to increase collision rates while large specks, or aggregates of collected inks, require lower turbulence to improve particle attachment and avoid further detachment from the air bubble. In general the experimental results were shown to be in agreement with the theory established in the field of mineral flotation regarding the effects of particle size, bubble size and turbulence on flotation kinetics. The trends observed in the experimental flotation cells were also consistent with deinking mill experience. As major recommendations, we suggest that ample air should be supplied to increase ink and speck removal efficiency, that aeration is achieved under high turbulent mixing conditions to produce small bubbles and promote collision with the microscopic ink particles, and that subsequent separation is achieved under lower and optimized turbulence in order to promote stable attachment of large particles onto bubbles and limit fiber losses. Strong agglomeration of the smallest ink particles may be another approach to optimize flotation conditions under lower turbulence. Current technology generally involves such flotation conditions, with some differences concerning relative optimization of aeration, turbulence and flotation time. In all

F. Julien Saint Amandr Int. J. Miner. Process. 56 (1999) 277–316

305

cases increasing the amount of small bubbles, turbulence and contact time leads to increased energy and investment costs to achieve the highest flotation efficiencies. Further improvement of flotation technology with respect to cost effectiveness is however expected on the basis of optimized hydrodynamic parameters such as turbulence, bubble size and contact-time distributions. These must be adapted to particle size, shape and density, as well as to surface chemistry in relation to the decisive microprocesses of ink agglomeration and liquid film drainage and rupture time.

Appendix A. Deinking flotation technology The evolution of the flotation units commercialized by the pulp and paper machinery suppliers shows a large diversity of designs ŽFig. 16. but also similarities in terms of the principles behind flotation deinking. Flotation cells or cyclones may be very different in size and shape, but the key elements ensuring aeration, ink collection and bubble separation are always present. In fact, ink collection under turbulent mixing conditions is always more or less included in the aeration and separation processes. This mixing process involves collision, attachment and unwanted detachment of ink particles and bubbles. Depending on the design of the flotation unit, this decisive step of the flotation process is mainly achieved within the aeration elements or within the flotation unit in connection to bubble separation. A.1. Aeration and collection Aeration refers to the introduction of air into the fiber suspension in a correct form and amount. Since the number and size of air bubbles crucially affect the probability of collision between bubbles and ink particles, the aeration conditions have to be characterized by relevant parameters such as bubble size distribution and volumetric air to pulp ratio ŽJulien Saint Amand and Perrin, 1991.. The air ratio, or air flow rate, can be defined at each aeration step if pulp and air are supplied through the same inlet, or globally for a flotation unit or a complete installation, on the basis of the cumulated air flow, which should be considered at the mean pressure within the flotation units. In fact the relevant parameter should be the effective air ratio, based on the air removed in the foam with the collected inks. However, the difference between removed and consumed air flow rate should be very small as generally only small bubbles, which do not contribute significantly to the air ratio, might not be removed in the foam. Increasing the amount of air effectively used in the flotation process, and more particularly the specific phase interface ŽKlein et al., 1994; Gottsching et al., 1995. ¨ involved in the collection of ink particles, may be regarded as a preferred means to improve ink removal, at each aeration step. However there are limits, depending on bubble size distribution and pulp consistency, to the amount of air which can be introduced into a fiber suspension in order to keep a sufficiently free motion of bubbles and fibers. As a general rule, larger amounts of air can be mixed in the pulp as the

306

F. Julien Saint Amandr Int. J. Miner. Process. 56 (1999) 277–316

Fig. 16. Current deinking flotation equipment. ŽA. LamortrFiberprep—Multi-aeration closed cell—Air injectors—three to five stages—Circular section ŽSerres and Colin, 1996.. ŽB. Beloit—Pressure deinking module—Pressurized air inlet—Static mixing zone—Cylindrical ŽMac Cool and Carroll, 1990.. ŽC. Voith Sulzer—Ecocell—Air injectors—Cells connected lengthwise with recirculation openings ŽKurz and Frymark, 1992.. ŽD. Black Clawson—IIM Flotator—Air manifold fed turbine—four sections lengthwise in one cell ŽGilkey and Yoshida, 1992; Gilkey et al., 1994.. ŽE. Kvaerner Hymac—Column flotation cell—Pressurized air Žor airrwater. spargers ŽDessureault et al., 1995.. Note: figures are redrawn from the literature cited as well as product literature.

bubble size is increased and a lower specific energy is required to create the interface. On the other hand large bubbles are less effective than small bubbles, but are easier to remove from the pulp suspension. Different aeration systems have been developed by the equipment suppliers. They include various air introduction systems, under pressure or by self suction, and hydrodynamic or mechanical mixing in the pulp flow, at the inlet or within the flotation cell. The mechanical method is based on a pumping rotor and generally stator elements, designed to achieve pulp circulation, air suction and mixing. This concept, developed and still widely used for mineral flotation, was the first to be adapted to deinking flotation but has now been dropped by the equipment suppliers who introduced it to the market. Other suppliers have developed the mechanical aeration method further ŽFig. 16D.. Air is injected under pressure through a manifold on a turbine in order to generate

F. Julien Saint Amandr Int. J. Miner. Process. 56 (1999) 277–316

307

and disperse the air bubbles and to provide multiple pulp recirculation within a long retention time cell ŽGilkey and Yoshida, 1992; Gilkey et al., 1994.. Several other methods of pressurized air introduction are currently used without mechanical dispersion: air is injected in the pulp flow through openings ŽFig. 16B. or porous media ŽFig. 16E. and subsequently dispersed under turbulent flow conditions ŽMac Cool and Carroll, 1990; Torregrossa et al., 1992; Dessureault et al., 1995; Carlton, 1996.. In the pressurized flotation technology ŽFig. 16B., separate zones are provided for the aeration step through openings in the feed pipe, and for the subsequent mixing step ŽMac Cool and Carroll, 1990.. The turbulence increase in the mixing zone is assumed to reduce the size of the bubbles produced in the aeration zone under lower turbulence and shear. Concerning the effect of pressurization, it must be pointed out that the pressure level within the flotation unit compared to the pressure variations in the aeration and mixing zone, should have little influence on bubble size distribution and on nucleation and growth of micro-bubbles as far as this mechanism is concerned. In cyclone flotation, initially developed for the mineral industry and recently adapted to deinking, air is sparged through the porous wall of a cleaner-type flotation unit, dispersed in the downward vortex flow and removed in the upward vortex core ŽTorregrossa et al., 1992.. The latest design of cyclone flotation has been modified in such a way as to increase the amount of air sparged in the flotation cleaner and entrained in the downward outlet flow in order to be further removed in a conventional flotation cell ŽCarlton, 1996.. Under these conditions the cyclone unit is essentially used to provide aeration and collection, separation within the cyclone being limited to the removal of excess air. The removal of excess air should correspond to large bubbles since the centrifugal separation effect has been reduced in the new cyclone design. A new centrifugal flotation process, using a rotating cell where air is introduced through a rotating perforated wall, has been developed recently on a laboratory scale ŽGrossmann et al., 1997.. Very high air flow rates, up to 50 times the pulp flow rate which is far beyond the air ratios of industrial flotation cells as reported in Table 3, can be achieved with this new technology, since the centrifugal flow field improves the separation of the air bubbles and can be maintained over a longer time than in cleaner type units used for conventional cyclone flotation. These advantages are also achieved with rotating cleaners ŽJulien Saint Amand et al., 1985. compared to conventional lightweight contaminant cleaners. In column flotation ŽFig. 16E., recently developed in the field of deinking and also adapted from mineral flotation where the technology showed to be very cost effective, the sparger system is arranged at the bottom of the column ŽDessureault et al., 1995.. Spargers made up of holes in which a mixture of air and water is injected and released in the cell, or porous spargers are used. A comparison of the various aeration systems using pressurized air supply, is not easy in terms of bubble size distribution. However it can be considered that smaller bubbles are produced as the air injection velocity Žor air flow rate. and the size of the pores or openings are decreased, and as the pulp-flow velocity in the air-injection area is increased. A decrease of the pulp surface tension also reduces bubble size. If turbulence is increased after air injection bubble size may be further decreased in the turbulent

308

Aeration method

Pressurized air

Reference wliteraturex

Carlton Ž1996.

Air ratio per aeration Ž%. Number of aeration steps Number of flotation units Total air ratio Ž%.

70–100 1 3 210–300

Air injector Mac Cool and Carroll Ž1990. 300 1 3 900

Gilkey and Yoshida Ž1992., Gilkey et al. Ž1994. 600–1000 Internal recirculation 1 600–1000

Serres and Colin Ž1996. 50 5 1 250

Britz Ž1993. 60 1 4 240

Pfalzer and Schweiss Ž1988., Kurz and Frymark Ž1992. 50 1 4–6 200–300

F. Julien Saint Amandr Int. J. Miner. Process. 56 (1999) 277–316

Table 3 Aeration conditions of current flotation installations

F. Julien Saint Amandr Int. J. Miner. Process. 56 (1999) 277–316

309

mixing zone. Under these conditions it is very difficult to evaluate, without measurements, bubble size distributions for such different air injection methods as, for example air sparged through very small pores in the rather low turbulence flow of column flotation ŽFig. 16E., and air blown through a much larger opening onto a high speed turbine ŽFig. 16D.. The hydrodynamic method, based on the self suction and mixing of air according to the Venturi principle ŽFig. 16A and C., has become the most common aeration technique since the first injector cells were specifically developed for deinking flotation ŽBarnscheidt, 1985; Barnscheidt, 1987; Serres and Colin, 1996.. Injectors of various sizes and designs have been investigated and constantly improved by the equipment suppliers and adapted to different designs of flotation units ŽPfalzer and Schweiss, 1988; Kurz and Frymark, 1992; Britz, 1993; Britz and Peschl, 1994; Chudacek et al., 1995; Carletti and Wood, 1996; Hori, 1996; Serres and Colin, 1996.. Referring to the pulp injection diameter, the size of the injector is regarded as the most important design parameter to control the pulp jet velocity, the suction effect, the surface of phase interface and the turbulence level within the injector, in relation to pressure drop and injector capacity. It is suggested that the hydraulic diameter Ž4 = arearperimeter. of pulp injection be used to define the injector size in the case of injector designs using numerous or non circular injection openings. Large injectors have been dropped and currently the injector flotation cells are equipped with various arrangements of small and medium size injectors, between 12 and 30 mm diameter ŽBarnscheidt, 1985; Britz, 1993; Carletti and Wood, 1996., or less in terms of hydraulic diameter ŽSerres and Colin, 1996.. The amount of induced air depends on injector size and design. The various possibilities applied to increase the air ratio include, a high jet velocity, a small hydraulic diameter, a large area around an unstable jet to promote air suction by suspended droplets, a low outlet pressure and pressurized air supply. High air flow rates up to 200%, and increased flotation efficiency on a pilot scale, have been reported under self suction conditions for a small injectors ŽGottsching et al., 1995.. However, the ¨ injectors used on industrial equipment are generally operated at a lower air ratio, in the range of 30–70%, since it was found by experiment that beyond certain limits of the ratio no further increase in the flotation effect was possible ŽPfalzer and Schweiss, 1988; Serres and Colin, 1996.. The size distribution of the bubbles produced by the injector depends mainly on the level of turbulence available to mix the air in the pulp, as well as on the air ratio and on the surfactants in the pulp suspension. A high turbulence is required to produce small bubbles. Such high turbulence is achieved by small injectors operated at high jet velocity. Injectors with turbulence generating devices in the mixing and distribution section have been developed in order to increase turbulence intensity and scale in such a way to produce air bubbles in a large size range ŽKurz and Frymark, 1992; Britz, 1993.. Depending on the design of the pulp injection nozzle the feed pressure is partly or almost completely transformed in kinetic energy to achieve the required injection velocity Žtheoretically 10–20 mrs for 0.5–2 b pressure reduction without friction losses.. Injection velocities of 9 mrs for about 1 bar feed pressure ŽKurz and Frymark, 1992. and about twice as much for a small injector fed at 2 bars ŽBarnscheidt, 1985.

310

F. Julien Saint Amandr Int. J. Miner. Process. 56 (1999) 277–316

have been reported. Much higher jet velocities of about 50 mrs have been reported recently for a particular high pressure injector using a thin, clear liquid jet to mix the air in the suspension ŽChudacek et al., 1995.. The effects of air ratio and liquid surface properties on bubble size distribution were investigated Žin the frame of previous, unpublished studies. in the case of a commercial 12-mm injector. Bubble sizes were measured on video frames taken from a transparent channel connected at the outlet of the injector. Clear water with and without surfactant as well as deinking process water were tested at different air flow rates set by a valve at the air inlet. Much smaller bubbles were obtained by adding surfactants or with process water. Bubble size was clearly decreased as the air flow was reduced and very small bubbles where observed at low air ratio. This result was assumed to be due to the large suction effect achieved as the air valve was progressively closed: as a vacuum is created around the suspension jet and the turbulence is assumed to be unaffected by the pressure level, less air is contained in a given bubble volume so that bubble size will decrease as the jet kinetic energy is recovered and the pressure increased. Recent investigations on different small injectors tested with water also showed the increase of bubble sizes with the air ratio ŽGottsching et al., 1995.. Bubble size ¨ distributions were measured by using a new optical method based on bubble volume measurement in a capillary. By contrast no significant variations of the bubble size distribution curves were observed with deinked pulp. The specific phase interface was thus proportional to the air ratio. However it must be pointed out that these results were obtained at high air ratios Ž50 to 200%. while our results on a single 12-mm injector were obtained in a lower range of less than 50% air. Concerning the microbubbles which may generated from dissolved air during pressure release, it has been suggested that they may be formed on ink particles and thus improve flotation efficiency ŽMac Cool and Carroll, 1990., while other authors reported, from experimental and theoretical considerations, that it is unlikely that new bubbles can be nucleated and grown on the surface of fibers or ink particles ŽAjersch and Pelton, 1993.. As already mentioned, aeration and collection are closely linked, because the collection process Žcollision and attachment between ink particles and air bubbles. begins as soon as air is mixed in the pulp, while the aeration process Žair introduction and bubble production. really ends as no more bubbles are produced in the pulp suspension. The technological approach to achieve the aeration and related collection steps, is thus of critical importance for the flotation efficiency. Current technology is compared in Table 2. The comparative data are based on typical deinking flotation installations recommended by equipment suppliers to achieve high removal efficiencies on detached ink particles under usual flotation conditions. The number of aeration steps or flotation units in series may vary according to waste paper grade, deinked pulp quality requirements and flotation consistency ŽBritz and Peschl, 1994.. Secondary flotation Žtreatment of flotation rejects., as well as post-flotation Žadditional flotation step implemented after a hot dispersion step to further detach ink particles from the fibers. which normally requires less aeration steps or flotation units in series, are not considered in Table 1. Under these conditions the total air ratio varies

F. Julien Saint Amandr Int. J. Miner. Process. 56 (1999) 277–316

311

between 200 and 300% for injector cells ŽBarnscheidt, 1987; Pfalzer and Schweiss, 1988; Kurz and Frymark, 1992; Britz, 1993; Britz and Peschl, 1994; Serres and Colin, 1996.. About three times more air is recommended for pressure aerated cells ŽMac Cool and Carroll, 1990; Gilkey and Yoshida, 1992; Gilkey et al., 1994., which is assumed to correspond to larger air bubbles. The values of total air ratio reported for the latest cyclone flotation technology Žcyclone mainly used for aeration and collection. are overestimated as the amount of vented air Ž25–50% of the air inlet flow. has not been considered ŽCarlton, 1996.. A.2. Collection and Separation High turbulent mixing in connection with pulp aeration at the inlet or within the cell is generally used to promote ink collection by increasing the probability of particle bubble collision. However these favorable conditions are only achieved over a very short time in order to keep the energy consumption within limits. Turbulence is then quickly decreased, as the pulp suspension moves to the separation zone. Additional ink collection is then achieved under much lower turbulence but over a longer time. The mechanisms of ink collection in terms of probabilities of attachment and detachment of ink particles with air bubbles are very complex. Ink collection is strongly affected by particle size, bubble size and turbulence intensity and scale. The mechanisms of ink separation in terms of stabilization and removal of the particle bubble aggregates are also strongly affected by hydrodynamic parameters which depend on the flotation technology. In the most common case of injector cells ŽFig. 16A and C., a major part of the ink to be removed with the injected air is collected onto the air bubbles in the injector and the feed area where the high turbulence level is progressively decreased. As the aerated pulp reaches the cell body, gravity collision between rising bubbles and ink particles becomes more significant compared to turbulent collision. The additional ink collection obtained within the cell body as turbulence decreases depends on the cell design and operating conditions. The parameters of decisive importance for gravity collision and bubble separation are the vertical flow component Žsuperficial velocity or surface related pulp flow. and the flow direction Župward or downward. which define the residence time and bubble particle contact time within the cell in connection with the pulp and air introduction levels. Unwanted short circuits and large-scale turbulent mixing may reduce the air removal efficiency if the residence time in the separation zone below the air introduction level is to short. Flotation cells with downward pulp flow and thus counter current with respect to the air flow are the most common. If aerated pulp is introduced at a single level ŽMac Cool and Carroll, 1990; Kurz and Frymark, 1992; Britz, 1993; Chudacek et al., 1995; Carlton, 1996. the cell is essentially used for bubble separation. If aerated pulp is introduced at different levels ŽBarnscheidt, 1985; Serres and Colin, 1996; Carletti and Wood, 1996. as shown in Fig. 16A, the rising distance for air bubbles from the lower injection levels to reach the surface is increased. The contact time for additional particle–bubble collision is increased, but the superficial velocity of the downward flow in the upper cell area has

312

F. Julien Saint Amandr Int. J. Miner. Process. 56 (1999) 277–316

to be decreased to compensate for the increased superficial velocity of the upward air flow. If the air is introduced in the cell at levels lower than the pulp suspension, such as in column flotation ŽFig. 16E., the average contact time is roughly proportional to the cell height for constant superficial velocities and bubble size distributions. The ink collection mechanisms in column flotation are exclusively governed by gravity mixing caused by the unstable rising motions of the air bubbles through the pulp flow. Large bubbles are most effective to produce some turbulence in the pulp suspension as they have higher buoyancy and rising velocity, but the contact time is lower. On the other hand small bubbles provide much longer contact time and may be entrained downward, since their relative upward velocity is much lower. Flotation cells with upward pulp flow have also been developed in order to increase the particle bubble contact time ŽBarnscheidt, 1985.. Under these conditions the migration distance of the air bubbles through the pulp is much lower for small bubbles having a low relative rising velocity, than for large bubbles although their residence time in the upward flow area is reduced ŽSerres and Colin, 1996.. As the flow configuration must be downward in the separation zone, upward flow is always connected to downward flow with intermediate horizontal pass. Such flow configuration is observed in different types of cells with multiple pulp aeration ŽBarnscheidt, 1985; Gilkey and Yoshida, 1992; Chudacek et al., 1995; Serres and Colin, 1996; Carletti and Wood, 1996. using internal ŽFig. 16D. or external recirculation ŽFig. 16A.. Among global engineering characteristics of flotation cells, specific power, retention time and superficial velocity are most relevant. According to mill experience ŽFloccia, 1995. and equipment suppliers ŽGilkey and Yoshida, 1992; Kurz and Frymark, 1992; Chudacek et al., 1995. the specific power consumed by a flotation installation to achieve high ink removal is generally in the range of 30 to 50 kW hrt, or down to 20 kW hrt according to recent results ŽSerres and Colin, 1996.. Trends to increase the installed power with respect to the cell volume have been pointed out and related to the turbulence observed at the surface of different cells ŽFloccia, 1995.. Values of specific installed power between 2 and 10 kW hrm3 were reported. The retention time in a flotation unit is directly related to the superficial pulp flow velocity, cell height and flow configuration. According to the type of cell and operating conditions, the retention time varies between about 20 s ŽLinck and Siewert, 1994. and 20 min ŽGilkey and Yoshida, 1992.. For complete flotation installations under usual deinking conditions as compared in Table 1, the cumulated retention time is in the range of 2–10 min. Low cumulated retention time is observed when ink collection is mainly achieved in the injector and the flotation cell essentially used to remove the air bubbles, while high retention time corresponds to additional ink collection within most of the cell volume. The superficial downward or upward flow velocities are in the range of 0.5–5 cmrs. Velocities between 1 and 2 cmrs are most common according to the values or connected data published by the equipment suppliers. In relation to these upward and downward surface velocities one can assess a cumulated distance of upward and downward flow which is the range of 3–10 m for a complete flotation installation. Small superficial velocities are necessary to remove small bubbles and to increase contact time in the case of column flotation and upward flow configurations. Conventional parallel

F. Julien Saint Amandr Int. J. Miner. Process. 56 (1999) 277–316

313

flow between inclined plates ŽChudacek et al., 1995. or special internal design ŽFig. 16A. in connection with very low surface velocity ŽSerres and Colin, 1996. helps to remove smaller bubbles. Depending on the large scale turbulent flow velocities in the cell, a minimum distance of about 1 m has to be kept under the aerated pulp introduction level in order to avoid turbulent diffusion of air bubbles which should be removed to the accept outlet. Depending on the foam removal technique Žpressure in Fig. 16A and B, overflow in Fig. 16C, scraber in Fig. 16D and E. and turbulence level in the cell, a minimum distance also has to be kept above the pulp introduction level, in order to selectively remove the inked bubbles. Some additional ink collection, as well as unwanted ink detachment and fiber collection, occurs in this foam removal zone. Numerous parameters such as foam height and stability, turbulence level, bubble size and air surface velocity are involved in the separation of the pulp components in this zone. Cyclone flotation is a special case in terms of residence time and bubble travel distance, where flotation is achieved within the cyclone ŽTorregrossa et al., 1992.. A high centripetal acceleration compared to gravity and a small travel distance depending on the cyclone diameter compensate for a very short residence time to enable the removal of small bubbles. With the rotating flotation cell ŽGrossmann et al., 1997., the residence time is increased while maintaining a high centripetal acceleration, which allows more small air bubbles to be used for particle collection. The working principle of the rotating cell copes with distinctly higher consistencies and leads to lower pulp losses, as observed with the rotating cleaner compared to conventional cleaners ŽJulien Saint Amand et al., 1985..

References Abrahamson, J., 1975. Collision rates of small particles in a vigorously turbulent fluid. Chemical Engineering Science 30, 1371–1379. Ahmed, N., Jameson, G.J., 1989. Flotation kinetics. Mineral Processing and Extractive Metallurgy Review 5, 77–99. Ajersch, M., Pelton, R., 1993. The growth of bubbles on pulp fibers and on carbon black dispersed in supersaturated carbon dioxide solutions. CPPA, 2nd Research Forum on Recycling, pp. 129–134. Ajersch, M., Pelton, R., 1995. Mechanisms of pulp loss in flotation deinking, 3rd Research Forum on Recycling, 20–22 Nov. 1995, Vancouver, pp. 27–40. Barnscheidt, W., 1985. Die Weiterentwichlung des Verticell Verfahrens. Wochenblatt fur ¨ Papier-fabrikation 113 Ž13., 484–485. Barnscheidt, W., 1987. Uberlegungen zur Auslegung eines Deinkinginjektors, PTS Deinking Symposium. Munich, p. 85. Bechstein, G., Unger, E., 1972. Ein Beitrag zur Klarung der Wirkungsweise von Sammlern beim Deinking ¨ flotationsprozess. Zellstoff und Papier 10, 297–306. Binder, G., 1993. La turbulence. ENSHMGrINPG course, Grenoble. Bird, R.B., Stewart, W.E., Lightfoot, E.N., 1960. Transport Phenomena, Wiley. Britz, H., 1993. Flotations deinking—Grundlagen und Systemeinbindung. Wochbl. Papierfab. 10, 394–401. Britz, H., Peschl, A., 1994. Flotationsdeinken bei erhohten Stoffdichten—oder wirtschaftlich flotiert. Wochbl. ¨ Papierfab. 15, 603–608. Carletti, C., Wood, R.J., 1996. Spidercel Reactor technology improves flotation deinking efficiency. TAPPI Recycling Symposium Proc., pp. 321–324.

314

F. Julien Saint Amandr Int. J. Miner. Process. 56 (1999) 277–316

Carlton, K.A., 1996. New deinking system combines GSC contacting with tower flotation technology. TAPPI Recycling Symposium Proc., pp. 325–331. Chudacek, M.W., Fichera, M.A., Estrina, H., Marshall, S.H., 1995. Initial experiments on newsprint deinking by the Fastflot process. Progress in Paper Recycling, pp. 26–31. Comolet, R., Bonnin, J., 1961. Mecanique experimentale des fluides. Masson et Cie. ´ ´ Davidson, J.F., Schuler, B.O.G., 1960. Bubble formation at an orifice in a viscous liquid. Trans. Int. Chem. Engrs. 38, 335. Dessureault, S., Levesque, M., Barbe, M.C., 1995. Column flotation: a new technology for deinking recycled pulp. TAPPI, Recycling Symposium Proc., pp. 251–255. Dorris, G.M., Page, ´ M., 1995. Deinking of toner printed papers: Part 1. Flotation kinetics, froth stability and fibre entrainment. 3rd Research Forum on Recycling. Vancouver, 20–22 Nov. 1995, pp. 215–225. Ferguson, L.D., 1991. A review of flotation deinking technology. Progress in Paper Recycling, pp. 17–23. Fischer, S., 1982. Modellvorstellungen fur in der Flotationsstufe des ¨ die physikalish-chemischen Vorgange ¨ Deinking-Prozesses. Wochenblatt fur ¨ Papier-fabrikation 110 Ž14., 511–514. Floccia, L., 1995. Some comments about industrial cells for DIP. International Wastepaper Technology Conference. London Gatwick, 13–15 June, 1995, p. 4. Fortier, A., 1967. Mecanique des Suspensions. Masson et Cie. ´ Galland, G., Bernard, E., Sauret, G., 1977. Aspect physico-chimique du desencrage. Revue ATIP, 31, pp. 10, ´ 374–378. Gilkey, M.W., Yoshida, H., 1992. A new high efficiency flotation device. Proceedings, 78th Annual meeting, Technical Section, CPPA, January 31, 1992, pp. B153–B158. Gilkey, M.W., Conte, T.O., Seifert, P., 1994. Field experiences with a new high air rate flotator. TAPPI, Recycling Symposium Proc., pp. 163–170. Gottsching, L., Hunold, M., Krauthauf, T., Muller, J., Putz, H.J., 1995. Effect of Air Volume and Air Bubble ¨ ¨ Size Distribution on Flotation in Injector Aerated Deinking Cells. CPPA, 3rd Research Forum on Recycling, 20–22 Nov. 1995. Vancouver, pp. 17–25. Grossmann, H., Hanecker, E., Schulze, G., 1997. Flotation of recycled pulp in a centrifugal force field. TAPPI Recycling Symposium Proc., Chicago, 14–16 April 1997, pp. 65–72. Heindel, T.J., 1997. The fundamentals of flotation deinking. TAPPI Pulping Conference Proceedings. San Francisco, 19–23 October 1997, pp. 521–533. Hinze, J.O., 1959. Turbulence. MacGraw-Hill. Hori, H., 1996. The present state of small scale papermaking in Japan. Treatment of waste paper. Japan Pulp and Paper 28 Ž4., 19. Hornfeck, K., 1982. Flotationshilfsmittel und deren Einfluss auf den Deinking Prozess. Wochenblatt fur ¨ Papierfabrikation 110–115, 542–544. Hou, M.J., Hui, S.H., 1993. Interfacial phenomena in deinking: Part 1. Stability of ink particle–air bubble aggregates in flotation deinking. TAPPI, Pulping Conference Proceedings, pp. 1125–1142. Ives, K.J., 1984. The Scientific Basis of Flotation. Martinus Nijhoff Publishers. Julien Saint Amand, F., 1991. Evaluation des techniques d’epuration en fonction des caracteristiques physiques ´ ´ des contaminants. Recherche de modeles. ` Revue ATIP 45 Ž4., 145–164. Julien Saint Amand, F., Perrin, B., 1991. The effect of particle size on ink and speck removal efficiency of the deinking steps. 1st Research Forum on Recycling. 29–31 October 1991, Toronto, pp. 39–47. Julien Saint Amand, F., Bernard, E., Lamort, P., 1985. Entwicklung eines rotierenden Cleaners fur ¨ hocheffizientz Leichtschmutzentfernung. Wochenblatt fur ¨ Papierfabrikation 113 Ž20., 779–784. Jordan, B.D., Popson, S.J., 1964. Measuring the concentration of residual ink in recycled newsprint. Journal of Pulp and Paper Science 20 Ž6., 161–167. Klein, R., Schwarze, D., Grossmann, H., 1994. Beitrag zur Berwertung des Einflusses der Luftblasengrosse ¨ auf das Deinkingergebnis. 6th PTS Deinking Symposium, 15–18 March, 1994. Kocer, H., Johnson, D.A., Thompson, E.V., 1995. Influence of air flow rate, time and total air volume on efficiency of toner removal in mixed office waste flotation. TAPPI, Recycling Symposium Proc., pp. 65–72. Kurz, E., Frymark, J., 1992. The voith multi injector flotation cell. Proc. Pan Pacific Pulp and Paper Technology Conf., Tokyo, 8–10 September 1992, pp. 8–10, 131–138.

F. Julien Saint Amandr Int. J. Miner. Process. 56 (1999) 277–316

315

Larson, A., Stenius, P., Strom, des Deinking Prozesses. Wochenblatt fur ¨ G., 1982. Zur Oberflachenchemie ¨ ¨ Papierfabrikation 110–114, 502–506. Laskowski, J.S., 1989. Frothing in Flotation. Gordon & Breach. Linck, E., Siewert, W.H., 1994. The new Escher Wyss flotation cell in an improved deinking system. TAPPI, Pulping Conference Proc., pp. 605–610. Mac Cool, M.A., Carroll, W.P., 1990. Pressurized deinking module. TAPPI, Pulping Conference, pp. 145–152. Mac Cool, M.A., Leblanc, P.E., 1989. Deinking and separation technology. PIRA Conference, Recent Developments in Waste Paper Processing and Use, Session 2, Paper 15. 28 Feb.–2 March, 1989, Gatwich. Mac Kinney, R., 1996. The rise and rise of flotation deinking. PPE, JanuaryrFebruary, 1996, pp. 21–23. Nguyen-Van, A., Kmet, S., Schulze, H.J., 1995. Collection events in flotation: the quantitative analysis of the particle–bubble collision and the attachment of particle onto bubble surface. 19th Int. Mineral Processing Congress, San Francisco, 22–27 October, 1995. Ortner, H., Wood, R.F., Gartemann, H., 1975. Deinking-Stand und Entwicklung. Wochenblatt fur ¨ Papierfabrikation 103 Ž6., 597–601. Pan, R., Bousfield, D.W., Thompson, E.V., 1992. Modelling particle–bubble dynamics and adhesion in air bubblersolid particlerliquid systems. TAPPI, Pulping Conference Proc., pp. 941–956. Pan, R., Paulsen, F.G., Johnson, D.A., Bousfield, D.W., Thompson, E.V., 1993. A global model for predicting flotation efficiencies: model results and experimental studies. TAPPI, Pulping Conference Proc., pp. 1155–1164. Pan, R., Paulsen, F.G., Kocer, H., Nerez, R., Johnson, D.A., Bonsfield, D.W., Thompson, E.V., 1994. A global model for predicting flotation efficiencies: Part II. Particle size and flotation rate predictions, and experimental studies and comparisons. TAPPI, Recycling Symposium Proc., pp. 291–301. Paulsen, F.G., Pan, R., Bousfield, D.W., Thompson, E.V., 1993. The dynamics of bubblerparticle approach and attachment during flotation and the influence of short-range nonhydrodynamic forces on disjoining film rupture. CPPA, 2nd Research Forum on Recycling, pp. 1–12. Pfalzer, L., Schweiss, P., 1988. Die Voith Multi-Injectorzelle-Fonktionsweise und Anwendung in einer modernen Deinkinganlage. Wochbl. Papierfab. 8, 303–307. Plate, H., Schulze, H.J., 1991. Modeling of the overall flotation process based on physico-chemical micro-processes—Technique and application. 17th Int. Mineral Processing Congress, Dresden, 23–28 September, 1991. Putz, H.J., Schaffrath, H.J., Gottsching, L., 1991. Deinking of oil and water-born printing inks—a new ¨ flotation deinking model. 1st Research Forum on Recycling, Conf. Proc., Toronto, pp. 183—190. Santos, A., Carre, ´ B., Roring, A., 1996. Contribution to a better understanding of the basic mechanisms involved in the pulping and flotation of offset ink particles. TAPPI, Recycling Symposium Proceeding, pp. 339–347. Schmidt, D.C., Berg, J.C., Feb. 1996. The effect of particle shape on the flotation of toner particles. Progress in Paper Recycling, pp. 67–77. Schulze, H.J., 1991. The fundamentals of flotation deinking in comparison to mineral flotation. 1st Research Forum on Recycling, Toronto, 29–31 October 1991, pp. 161–167. Schulze, H.J., 1992. Probability of particle attachment on gas bubbles by sliding. Advances in Colloid and Interface Science 40, 283–305. Schulze, H.J., 1994. Zur Hydrodynamik der Flotations-Elementarvorgange. PTS Deinking Symposium, 5–8 ¨ May 1992. Wochenblatt fur ¨ Papierfabrikation 5, Munich, 1994. Schulze, H.J., 1996. Personal communication. Schweizer, G., 1965. Wochenblatt fur ¨ Papierfabrikation 93 Ž19., 823–830. Serres, A., Colin, E., 1996. The MAC flotation cell on a new deinking concept: Theory and practical results. 7th PTS Deinking Symposium, Munich, 7–9 May, 1996. Stratton, R.A., 1992. The surface chemistry of flotation of stickies and laser printed inks. Journal of Pulp and Paper Science 18 Ž6., 221–224. Torregrossa, L.A., Chamblee, J.W., Platner, G., 1992. Flotation deinking with the gas sparged cyclone. Pan-Pacific Pulp and Paper Technology Conference, Tokyo, 8–10 September 1992, pp. 101–104. Vidotti, R.M., Johnson, D.A., Thompson, E.V., 1995. Hydrodynamic particle volume and its relationship to mixed office waste paper flotation efficiency: Part. 1. TAPPI, Pulping Conference Proc., pp. 203–213.

316

F. Julien Saint Amandr Int. J. Miner. Process. 56 (1999) 277–316

Weiss, T., Schubert, H., 1987. Zum Einfluss des Feinstkornes auf die Turbulenz der Mehrphasen Stromung ¨ festrflussig in einer Flotationszelle. Kolloquim Mehrphasenstromung, Antonshohe, ¨ ¨ Report no. 16, pp. 133–147. Yoon, R.H., 1991. Hydrodynamic and surface forces in bubble–particle interactions. Aufbereitungs-Technik 32 Ž9., 474–485.