Table of Contents
The development of solvent extraction processes over recent years has led to their application on an increasing scale in metal extraction plants. The author’s company has recently been awarded a contract for the design of the world’s largest solvent extraction plant based on pump-mix mixer-settlers. The plant is designed to extract copper from a leach liquor and has an aqueous throughput of 12,000 imperial gpm. The design is based largely upon data obtained from the company’s own research laboratories, where work has been undertaken on mixer impeller design, baffle design, settler area, organic entrainment recovery, etc. Investigations have been conducted into the effects of mixer impeller speed, recirculation rates, phase continuity, phase ratio, etc., on extraction efficiency, settler area, and organic entrainment losses. The implications of the results of this work on the design of a large plant using pump-mix mixer-settlers are discussed and the techniques used in overcoming certain scaling problem are outlined.
The authors company has recently undertaken a development program aimed at exploring the relationships between the variables of greatest importance in the design of mixer-settlers for metallurgical solvent extraction applications. Such applications vary from relaively small-scale plants treating copper mine waters, nickel-containing effluents, etc., where the total flow rates may be of the order of a few hundreds of gallons per minute, to very large-scale plants treating several thousands of gallons per minute of leach liquors containing copper, uranium, or other metals.
This paper outlines the physical and mechanical factors which must be taken into account when designing mixer-settler systems for large-scale solvent extraction plants. The performance of the equipment will also be dependent upon the chemistry of the system, including the kinetics and thermodynamics of the ion exchange reaction itself, the effect of impurities, flocculants, surfactants, temperature, etc., but these will not be dealt with here.
Much of the laboratory work has been carried out using copper-containing solutions and either LIX 64 or LIX 64N as the solvent dissolved in various grades of kerosene. For simplicity, all the graphs and tables of results which are presented and discussed here will refer to the LIX-copper system, although the findings can be used as a basis for the evaluation of other systems. The LIX-copper process is now sufficiently well known to obviate the need for a full description; however, an outline flowsheet of the process is shown in Fig. 1.
The ultimate objective of a laboratory program for a particular extraction application is to provide data upon which a mixer-settler system can be designed to handle a given throughput of leach liquor and extract from this liquor the maximum quantity of metal for the minimum cost. In other words, the objective is to determine the design and operating parameters which achieve the optimum balance between capital and operating costs.
In order to achieve such an objective it is necessary to work not only with the particular system concerned, say LIX-copper, but also with solutions having the same pH, concentrations, impurities, etc., as will be experienced on the commercial plant. The most economic design cannot be arrived at solely by reference to the extensive literature which generally deals with different applications, nor by reference to past experience with other systems, although naturally this is always helpful.
A brief summary of some of the published work of most interest in this field is included first to provide the background against which the development work was undertaken. It should, however, be stressed that there is no substitute for experimental work carried out in the laboratory on one’s own particular solutions if an optimum plant design is to be achieved.
Review on Some Past solvent extraction Work
There is a growing body of literature developing on the fundamental aspects of droplet formation, mass transfer, and coalescence in two-phase liquid-liquid systems such as those commonly encountered in solvent extraction processes. In the U.K. alone the work of Davies and Jeffreys’- at Manchester and Aston, Hartland” at Nottingham, and Sawistowski at Imperial College is all aimed at obtaining a thorough understanding of the basic mechanisms at work in the solvent extraction process.
In addition to the more fundamental work being carried out in the Universities there is a well established body of knowledge on mixers and mixing processes. Much of this work has been undertaken by commercial concerns which have developed and marketed their own designs of mixing equipment. The work of Treybal is probably the best known and is one of the standard textbooks on the subject. Rushton, Costich, and Everett have investigated the power characteristics of various mixer impellers as have Miller and Mann. Conolly and Winter and Ryon, Daley, and Lowrie are amongst the many authors who have investigated the problems of scaling- up mixing operations.
Finally there is the applied work which has been done in connection with the development or operation of mixer settlers for specific duties. Since the large-scale
commercial applications of solvent extraction generally have been confined to the petroleum and uranium industries, most of the early work was carried out on these systems, and much of it was restricted information by virtue of its connection with the atomic energy programs in the UK and US. Colven, Lowes and Larkin, Coplan, Davidson, and Zebroski, Webster and Williamson, and Hanson have all published works on particular mixer-settler designs and Treybal has presented a mathematical approach in considering the factors affecting the economic design of mixer-settlers. Surveys of contacting equipment available in the market have been made by Hanson and Akell.
Factors Affecting the Cost and Performance of a Mixer-Settler
In order to arrive at the most economic design of a mixer-settler system for a given application, it is necessary to consider all those factors which affect both the capital and operating costs of the system.
The factors having a significant effect on the capital cost are: 1) number of stages of mixer-settlers, 2) settler area per stage, 3) mixer volume per stage, 4) solvent inventory, 5) solvent recovery equipment, and 6) ancillary equipment including pumps, pipework, instrumentation, etc.
Similarly the operating cost is made up of: 1) cost of solvent loss, 2) cost of acid loss, 3) cost of power to mixing, 4) cost of power to pumping, and 5) operating labor cost.
In addition, the economics of the process are greatly affected by the extraction efficiency achieved, which determines the value of unextracted metal in the raffinate.
Most of the variables which affect the capital and operating costs of the plant rely upon the relationships between the degree of mixing to which the two phases are subjected, the extent of mass transfer which occurs, and the degree of phase separation which subsequently can be achieved.
Thus, ideally the laboratory program would be designed to investigate the relationships between mixing, mass transfer, and phase separation in the many possible configurations of mixer-settler and over the whole range of operating conditions.
Since the ultimate purpose of the contacting operation is to achieve mass transfer of metal from the aqueous phase to the organic, and vice versa, it is convenient to use the extent to which mass transfer occurs as the dependent variable in assessing mixer performance and to measure against this the effect of variations in the other, independent variables.
The term used to describe the extent to which the mass transfer process has reached completion is the “percentage approach to equilibrium” which is defined as the ratio (expressed as a percentage) of the actual aqueous extraction efficiency achieved in a mixer-settler operation (either single or multistage) to the theoretical aqueous extraction efficiency achievable with the solutions in question, as determined from the appropriate McCabe-Thiele diagram. However, it is convenient to refer to the “percentage approach to equilibrium” as the “extraction efficiency” and this latter term will be used in the text of the paper.
In order to assess the performance of the settler it is necessary to consider two dependent variables, namely: 1) aqueous entrainment (ppm) in the organic solution leaving the settler; 2) organic entrainment (ppm) in the aqueous solution leaving the settler.
There are two distinct facets of settler performance. First, the settler must be of such size as to ensure the substantially complete separation of the mixed phases for the given flow rate. In other words, flooding must not occur. The specific settler flow factor determines whether this criterion will be met and this depends largely on the chemical and physical characteristics of the phases involved. However, as will be described later, the factor is also affected by mixer performance. Second, even though the settler is designed to prevent flooding, the two phases leaving a settler in practice are never completely separated and each phase contains trace quantities of the other phase finely dispersed throughout its bulk.
Many of the factors which determine the total performance of the settler with respect to both facets are the same as those which affect the mass transfer performance of the mixer.
The independent variables (physical and mechanical), whose effect on the mass transfer and phase separation operations it is necessary to determine, are:
- mixer configuration,
- agitator speed,
- residence time in mixer,
- phase continuity,
- organic/aqueous ratio,
- specific settler flow, and
- settler configuration.
These affect the flow patterns and velocities in the mixing and settling tanks, and although the list is not exhaustive, it includes those variables of most practical importance in the design of a solvent extraction system.
It is clear that both the mixer and settler performance could be expressed in terms of the fundamental properties (droplet size, interfacial tension, etc.) of the mixed phases produced by the mixer and fed to the settler, but for the purpose of producing an optimum design with respect to mass transfer on one side and solvent inventory and loss on the other, it is necessary, for the LIX-copper system, to link the two unit operations and assess their combined effect.
Each of the independent variables itemized previously will now be considered separately and the effect of alterations to them on the performance of the mixer-settler will be discussed.
The configuration of the mixer includes not only the design and location of the impeller in the mixing chamber, but it also includes the shape of the mixing chamber, the location of the inlet and outlet, and the design and location of baffles.
All these should be viewed together as constituting the mixing environment since they all profoundly affect the velocities and flow patterns developed in the mixer and hence the droplet size distribution produced. For the same reason it is necessary to maintain geometric similarity between the laboratory and the ultimate plant scale if reliable scale-up of mixer performance is to be achieved.
Thus in determining the mixer configuration to be used in the laboratory test work, consideration must be given to the implications of a geometrically similar large scale plant. As a mixer-settler is scaled-up, its overall design alters as is shown in Fig. 2.
Assuming a typical retention time of 2 min and settler area of 2 US gal per sq ft per min, then the sizes of bench, pilot, and commercial plant units may be similar to those given in Fig. 2.
Two particular disadvantages of the design shown in Fig. 2c are worthy of mention, in addition to the limitation in the size of mixer that it is practical to operate. First, there is no longer full-width flow from the mixer into the settler and there is consequently a tendency for maldistribution of flow across the settler, leading to inefficient settling. This may be partially overcome by the installation of a suitable baffle at the inlet to the settler which distributes the mixed phases across the full width of the settler. More will be mentioned of this in the section on settler design.
Second, the adherence to the geometric scale-up criterion results in the mixing chamber gradually becoming deeper until at a total throughput of 10,000 US gpm, it has a depth of 13.9 ft. Since the organic phase in the weir of the previous settler has a specific gravity of, say, 0.8 and the mixed phases in the mixing chamber have a mean specific gravity of about 0.9, the static head to be overcome in order to pump the organic phase into the mixing chamber increases with increasing mixer depth (see Fig. 3). For a 14-ft-deep mixer this head difference is almost 1½ ft of water, a significant proportion of the total head available if the mixers themselves provide the pumping head. A further factor to be taken into account when determining the depth of the mixer is that the settler, generally only 2-ft 5-in. deep, must have the same top level as that of the mixer if hydraulic head is to be conserved, and this means that the settler has to be supported on columns or a plinth above ground level. The alternative of locating the mixers in a trench or pit is not attractive from an operating or maintenance point of view.
It appears from experience on some plants that phase instability is greater on large mixers than on small ones and phase continuity inversion has been reported to occur with relatively small O/A ratio changes. This may be indicative of the poorer mixing obtained in a large mixer where dead areas can occur, leading to large local variations in O/A ratio and consequent inversion. The importance of phase inversion is discussed later.
The choice between cylindrical and cubic mixing tanks is one which depends more on construction considerations than on effectiveness of mixing, which can be good in both cases. If the mixer-settlers in the full-scale plant are to be constructed in steel, then cylindrical tanks are the more economical design, whereas reinforced concrete construction favors the cubic design. With a cylindrical design side wall baffles are essential if vortexing is to be prevented and good mixing obtained, but in the cubic design the baffling effect is automatically achieved by the rectangular shape of the mixing chamber.
Next, there is the choice of impeller to consider. In this there is the widest scope for variation in design, from the axial flow propeller type to the radial flow flat-bladed paddle, either with or without shrouds.
The experimental work referred to here was largely confined to types of impeller which were potentially capable of providing the pumping head required in the full-scale plant.
The reasons for choosing a “pump-mix” impeller were twofold. The obvious saving in capital which can be achieved if the mixer is also capable of providing the necessary interstage pumping becomes more significant as the plant throughput is raised. Consider a plant designed to handle a combined throughput of 12,000 US gpm (6,000 US gpm organic and 6000 US gpm aqueous solution) in four stages of extraction and two stages of stripping. By using pump-mixers, eight 6000- gpm or, more probably, sixteen 3000-gpm pumps can be eliminated.
However there is another disadvantage of interstage pumps which may be more significant than the added capital cost. Each separated phase passing from one stage to the next invariably contains a quantity, generally small, of the other phase dispersed throughout it. The quantity of entrainment of one phase in the other will depend on a number of factors, but the size of the entrained droplets can be greatly reduced by the high degree of shear which occurs during passage through a pump; this may lead to subsequent settling difficulties and high levels of solvent loss and acid carryover.
The problem of high shear in the mixer also causing the formation of very small droplets and resulting in high solvent losses was one of the most important factors affecting the choice of impeller. A number of plants operating in the metallurgical field were known to have relatively high solvent losses and an effort was made to design a mixer which would overcome this problem.
The combination of the requirement for a pump-mix impeller and for an inherently low-shear design led to the choice of an impeller basically similar to that used in some of the early atomic energy work. A large number of variations of tank/impeller diameter, shroud clearance, and blading design were used in an attempt to obtain good mixing at the same time as a satisfactory pumping characteristic. The design shown in Fig. 4 was then used in a series of experiments to measure flow, head, power, and extraction efficiency under a variety of impeller conditions.
In the early work the draft tube was fixed to the impeller and was rotated with the impeller, so that there was a clearance fit between the bottom of the draft tube and the hole in the base plate of the mixing chamber.
Fig. 5 illustrates how the specific head/specific flow curves vary as the clearance between the draft tube and bottom plate is increased, and the extent of mixed phase recirculation is consequently increased.
In all the work on impeller flow, head, and power characteristics the nondimensional form of these functions has been used to relate results obtained on geo-metrically similar impellers (including mixing tank similarity) with different diameters and different speeds, as is common practice in pump design. This technique has produced consistent results over the range studied, within the limits of experimental error.
It can be seen clearly from Fig. 5 how the pumping head produced by a given impeller can be substantially reduced by increasing the recirculation rate in the mixing chamber. However, Fig. 6 illustrates the benefit that can be obtained by using increased recirculation to improve extraction efficiency.
Thus if a plant is overdesigned with respect to the head that the impellers can produce, the extent of recirculation, can be increased and the extraction efficiency improved as the head is reduced. The alternative of partially closing a butterfly valve in the feed pipe or draft tube to increase the pressure drop in the line has no beneficial effect on the extraction efficiency and therefore is wasteful of power. Similarly, reducing the impeller speed and thus reducing the pumping head also reduces the extraction efficiency which is generally unacceptable.
The initial design of impeller was very successful from the point of view of reducing the organic entrainment in the laboratory models and a series of experiments was conducted on impellers with various designs of shear blades attached to the top and bottom shrouds, Fig. 7. This series of experiments was designed to show how extraction efficiency (Fig. 8) and entrainment (Fig. 9) varied with increasing shear blading and speed so that the optimum speed and impeller configuration could be determined.
From the results two important conclusions can be drawn. First, that the extraction efficiencies of all the various impellers increase with increasing speed until they appear to reach a maximum which they approach asymptotically, Fig. 8. Second, Fig. 9 shows that aqueous entrainment is significantly higher for all impellers with shear blades than it is for impellers without such blades, operating under identical process conditions. For organic entrainment, no such trend is observed but organic entrainment does increase noticeably with increased impeller tip speed. The graph shows aqueous entrainment as a scattered zone since there was no consistent trend of increasing entrainment values with either speed or increasing shear blading. Before the choice of impeller can finally be made it is necessary to know the head, flow, and power characteristics of each design and these are shown in Figs 10 and 11.
From the graphs it can be seen that the pumping head developed increases consistently with the addition of larger shear blades, while the power absorbed is dramatically increased. All designs of impeller are capable of pumping the head required and so the economics depend on achieving a balance between power consumption, extraction efficiency, and entrainment.
Such a balance depends on the system being considered, its sensitivity to alterations in stage efficiency, the cost of the solvent losses, the value of unextracted metal, and the cost of power.
Finally, the vertical location of the impeller in the mixing tank must be determined. In the case of the double shrouded impeller design shown in Fig. 4 the presence of the draft tube enables the vertical location of the impeller to be varied independently both of the pumping head produced and of the extent of internal recirculation within the mixer. Whereas in the single shrouded design the impeller runs near the base of mixing tank which acts as its bottom shroud. In this design the vertical distance between the impeller and the base of the tank determines not only the flow pattern in the mixer, but also the pumping head and recirculation produced at any given impeller speed.
In locating the double shrouded impeller within the mixing tank it must be remembered that the feed to the impeller is a positive one through the draft tube and that, if the outlet of the mixer is through the top baffle (Fig. 4), then the fluid in the base of the mixing tank will be poorly mixed at low impeller speeds and settling of the heavy phase will occur. This effect can be eliminated if the draft tube is attached to the impeller and recirculation occurs at the base of the tube. However it is generally undesirable to have the draft tube integral with the impeller since it adds to the length and weight of the impeller, and a preferable arrangement is to provide a few small recirculation holes in the base of the draft tube which can then be fixed to the bottom of the mixing tank.
Miller and Mann have shown that the mixing efficiency in a two-phase system depends on the location of the impeller relative to the static interface level in the mixing chamber. They concluded from tests with a variety of impellers that the impeller should always operate in the heavy phase zone of a two-phase system and that the optimum height is either 0.4 times the total liquid depth from the bottom, or the interface height less the impeller width, whichever is less. In the same paper, however, Miller and Mann indicate that variations in the ratio of impeller depth to tank depth within the range 0.1 to 0.75 do not affect the power consumption significantly.
The effect of an increase in impeller speed on extraction efficiency, dispersion band thickness (depth of uncoalesced mixed phase in the settler), and entrainment has already been mentioned.
The results for all mixers show the same tendency for the extraction efficiency to increase with increasing speed until it approaches a maximum value.
The maximum value is achieved at different speeds for the different impeller designs and appears to be dependent upon the establishment of certain flow patterns and flow velocities in the mixer. Beyond this point the limitation in extraction efficiency is not one of mixing but is more probably related to the kinetics or thermodynamics of the reaction.
The fact that 100% approach to equilibrium was not achieved in the tests shown in Fig. 8 is due to the fact that the solutions used in the tests were purposely chosen to approximate to the last stage of a four or five-stage extraction. This stage is very sensitive to mixing efficiency and residence time in the mixer since the rate of mass transfer in it is lower than in any other stage due to the low “driving force” which exists.
From the graphs it appears that the design of mixer does not in itself greatly affect the ultimate extraction efficiency that can be obtained, but that it is most important to operate the chosen design at or above a certain minimum speed which gives a close approach to the maximum extraction efficiency that can be obtained. The speed and the maximum extraction efficiency are both dependent on the chemical system being considered and even on the stage considered within a multistage system. In practice, when dealing with a multistage system one generally would design for the mixer conditions necessary to obtain the maximum extraction efficiency in the last stage on the basis that that is the most exacting duty.
The choice of impeller speed is also affected by the variation of entrainment with speed. The results shown in Fig. 9 were obtained for various impeller designs and do not indicate a consistent trend. The inconsistencies may be due to experimental error but they probably also reflect a similar inherent unrepeatability that is experienced when measuring coalescence rates. There is reason to suppose from experience in a number of plants that the extent of entrainment of the dispersed phase increases with the tip speed of the impeller, and that it is desirable to minimize the impeller speed for this reason.
Dispersion band thickness has been shown to increase with impeller speed and hence the specific settler flow required in order to maintain the same dispersion band depth decreases with increasing speed. Again, therefore, it is advantageous to minimize the impeller speed in order to minimize the capital cost of the settler.
Experiments with some systems have indicated however, that the dispersion band depth may not always increase with impeller speed over a wide range of operating values, though it is always difficult to eliminate entirely the effect of temperature and other variables.
All these observations are, of course, related to a given impeller design. There is not necessarily any relationship between the effects of one design of impeller at a given speed and the effects of another design at the same speed. It is therefore very misleading to talk of critical impeller tip speeds, above which entrainment becomes unacceptable, since this depends on many other factors including the chemical and mechanical environment in which the impeller is operating as well as the design of the impeller itself. Figures of maximum allowable tip speeds should therefore be treated with caution since they generally refer to a certain specific impeller design in a specific organic/aqueous system.
The residence time required to achieve a high approach to equilibrium in a mixer is again dependent on the solutions being used, the mixer configuration, etc. In LIX-copper systems 2-min residence times have been conventionally used in both the extraction and stripping stages. The experimental results obtained show that higher approaches to equilibrium are achievable in a single-stage operation if 3-min residence time is used instead of 2 min. However the increased overall extraction efficiency obtained in a multi-stage unit is only very slightly increased by the 2-3% increase in approach to equilibrium per stage.
Thus the optimum residence time must be considered in relation to the overall extraction efficiency rather than individual efficiencies, though the latter provide a good guide to the values worth investigating on a multistage rig. The same comment applies to all the factors which affect the approach to equilibrium in a single stage.
Apart from the mean residence time which is determined by the relationship of the mixer volume to the total flow-rate through the mixer, the distribution of individual droplet residence times is also important when considering mixer extraction efficiency. The droplet residence time distribution is governed by the flow patterns within the mixer, which themselves are governed by the location of the inlet, outlet, and impeller. The positive feed of the incoming fluids to the eye of the impeller through a draft tube and the inclusion of a top baffle shown in Fig. 4 are both factors which assist in reducing the scatter of droplet residence times about the mean value. Although actual measurements of droplet residence times have not been made in the testwork reported here, an improvement in extraction efficiency is achieved by the inclusion of the top baffle with a central outlet. The improvement in efficiency is explained by the effect the baffle has in reducing the bypassing which tends to occur where the outlet is a simple weir on one side of the mixer.
The residence time of the mixed phases in the dispersion band in the settler may also contribute to the overall mass transfer observed across a mixer-settler and care must be taken to assess the extent to which this occurs before scale-up is undertaken. For example, if the specific settler flow is 2.4 US gpm per sq ft and the dispersion band depth is 4 in., then the mean residence time of the mixed phase in the dispersion band is 1 min, which is significant when compared with a mean residence time of 2 min in the mixer. In tests carried out to date with the LIX-copper system, the variation of mixer retention time in the range 1 to 4 min has appeared to have no significant effect on entrainment.
The importance of phase continuity will be discussed here from the physical and not the chemical point of view.
The extraction efficiency obtainable in a given mixer-settler can be significantly affected by varying which phase is continuous and which is dispersed, though the difference is less noticeable at high impeller speeds where the extraction efficiency is high for both phase continuities. For example in one test at an N³D² of 50 with internal mixer recirculation organic continuous operation gave a 91% approach to equilibrium while aqueous continuous operation gave 72% approach to equilibrium with the same solutions. This large difference in percentage approach to equilibrium could be explained partly by the thicker dispersion band obtained under organic continuous operation leading to a greater effective contact time for the two phases. However, the effect of phase continuity and direction of mass transfer on the particle size and mixing pattern in the mixer may also be significant, and further work will be undertaken to investigate this.
Dispersion band thickness in the extraction settlers are substantially greater when the mixer is operating with an organic continuous dispersion than they are when the aqueous phase is continuous (Fig. 12). In the stripping section, however, where the copper mass transfer direction is reversed and the acid concentration is higher, the dispersion band depths under both aqueous and organic continuous conditions are very much less than in the extraction section and there appears to be little difference in the depths characteristic of the two phase continuities.
It is common practice to run a mixer-settler organic continuous if the organic entrainment is to be kept low. That this philosophy is correct is born out by the curves shown in Fig. 9. The very low entrainment of the continuous phase which is experienced implies that most entrainment is caused by the production of very fine droplets of the dispersed phase due to high shear in the mixer. The fact that there is still a very small amount of entrainment of the continuous phase in the other phase leaving the settler is evidence of the fact that true “secondary haze” produced during coalescence in the settler also occurs, but this is very much the lesser of the two effects.
The marked decrease in entrainment of the organic phase if the mixer is operated with the organic phase continuous makes it desirable to run the last stage of
extraction (raffinate) and the first stage of stripping (advance electrolyte) organic continuous to minimize solvent loss and organic carryover to the electrowinning cells. Similarly the acid carryover from the stripping circuit to the last stage of extraction, which can result in low extraction efficiency and direct copper and acid losses to the raffinate, can be reduced if the last stage of stripping (spent electrolyte) is operated aqueous continuous. The first stage of extraction may also be operated aqueous continuous but the effect of leach liquor carryover on the acid and impurity levels in the electrolyte circuit is small and generally not economically significant.
In practice it is sometimes difficult to achieve and maintain the desired continuity in large plant mixers and frequently all the mixers are operated with the same phase continuity, which for the LIX-copper system is generally organic continuity since solvent losses are the dominant consideration. Surges of one phase or the other due to flow control instabilities or weir level instabilities may occur and cause inversion progressively through the system and the desired continuity can only be regained by stopping the mixer and restarting, or by isolating the flow of the phase required to be dispersed until the desired phase continuity is achieved.
The stability of each type of dispersion is a factor which has to be considered in the design of a mixer- settler. With a 1:1 organic to aqueous (O/A) ratio the phase which becomes continuous is generally the one in which the impeller is situated when it is started up. However, as the O/A ratio is made progressively greater or less than unity, then the tendency is naturally for the phase which is present in the largest quantity to become the continuous phase. There are other factors to be taken into account when considering the stability of phase continuity and these are the degree of mixing and the materials of construction of the mixer. Both these factors are important since they affect local O/A ratios which may differ from the mean value of the O/A ratio in the mixer.
Poor mixing may result in localized settling which can alter the phase continuity in parts of the mixing tank. Such alterations can quickly spread throughout the mixing tank if the phase continuity is already near to its inversion point. Similarly the use of organophilic materials (epoxy resins, perspex, etc.) in the construction of the mixer can cause a preference for organic continuity to occur in the vicinity of these materials, though this effect usually only has significance in small mixers where wall effects are important.
The phase ratio within a mixer may be adjusted to a value such that the desired phase continuity is obtained despite an unfavourable overall phase ratio by recycling a quantity of one or other phase from the settler outlet weir back to the mixer.
The organic/aqueous ratio (O/A ratio) is not generally considered in relation to the extraction efficiency in the mixer, the specific flow in the settler, or indeed to the entrainment experienced at the settler outlet, though it undoubtedly does have some effect upon these functions. The O/A ratio is normally fixed at a value which ensures that the desired phase continuity is achieved in the mixer. At 1:1 O/A ratio either phase continuity can be achieved depending upon which phase is covering the impeller when it is started up.
The O/A ratio beyond which it is not practical to achieve organic continuity in a mixer is considered to be about ¼ or 1/5, though theoretically by considering the fluid bulk being filled with closed packed spherical droplets this ratio would be expected to be 1/3.5. The reverse is true for aqueous continuity, though it is found that some systems appear to exhibit a preference for one continuity or another.
The other important effect of phase ratio is that it governs the overall concentration ratio which can be achieved across a solvent extraction system. Conventionally the phase ratio in the extraction circuit is set at 1:1 and that in the stripping circuit is such that the copper drop across the electrowinning tankhouse is at the desired value. However, recycle is generally used to alter the O/A ratio in individual mixers in the stripping circuit to 1:1 or as close to that value as possible. The 1:1 O/A ratio not only enables either phase continuity to be achieved relatively easily, but it is also the ratio at which the mixing efficiency of a given system is a maximum.
Another effect of altering the phase ratio in the extraction circuit is to alter the amount of copper to be extracted by a given quantity of solvent. This then alters the percentage of the solvent required in the diluent (kerosene) in order to obtain the required recovery. Alteration of the phase ratio does not, however, affect the final pH of the raffinate since that is solely dependent upon the pH of the feed liquor and the quantity of copper extracted.
From the foregoing it can be seen that the phase ratio chosen for each of the circuits does have a marked effect on the capital cost of plant since it determines the flow rates to be used in the organic and electrolyte circuits, and also the solvent concentration and hence the settler area required. For example, the flow to the tankhouse can be halved and the copper drop across it doubled by doubling the overall O/A ratio in the stripping mixer settlers.
Specific Settler Flow
The settler cost is proportional to the specific settler flow and its performance is judged by the values of entrainment in the effluent streams.
The dispersion band depth is a function of both the specific settler flow and the phase continuity as is shown by the curves in Fig. 12. Since there is always the possibility of phase inversion, even if aqueous continuity is the desired mode of operation, the settler must be designed with sufficient area to prevent flooding under organic continuous operation. A margin also should be allowed in the design in order to take into account any blinding of the coalescing interface which may occur due to solids or fungus collecting in the dispersion band.
A further point which should be noted in considering Fig. 12 is the dependence of specific settler flow on the performance of the mixer, as represented by the N³D² factor. For the LIX-copper system it has been clearly shown that increase of mixer speed produces dispersions of poorer settling properties so that, although increased mixer speed can lead to better mass transfer, the total benefit can be diminished by the effect of increased settler size on capital equipment costs and solvent inventory.
Temperature greatly affects the rate of coalescence and hence the dispersion band depth and it is necessary to design for the lowest nighttime temperature at the site if settler flooding is to be eliminated. Flooding problems, if they exist, may be overcome by installing heaters in the plant to heat the feedstock to, say, 70°F, at which temperature the rate of coalescence should be sufficient to prevent flooding.
Solution concentrations also have a significant effect on the relationship between specific settler flow and dispersion band depth. As the concentration of the solvent in the kerosene carrier is increased, the dispersion band depth increases until it eventually limits the percentage of solvent with which it is possible to operate. The phase disengagement properties of high concentration solvent solutions can occasionally provide a limitation on the quantity of metal which it is possible to extract from a given liquor. Higher acid concentrations generally improve the rate of coalescence and reduce dispersion band depths. It is possible, therefore, to design the stripping settlers for a higher specific flow than the extraction settlers in order to achieve similar dispersion band depths.
Having taken these factors into consideration the settler area required for a particular duty can still be minimized by good design.
First, the method of introducing the mixed phases into the settler is important It has been shown by Davies and Jeffreys that, by varying the level at which the mixed phase inlet to the settler is located, the dispersion band depth can be varied for a given throughput, the minimum dispersion band depth being achieved when the inlet is into the dispersion band at the coalescing interface. It is also advantageous to minimize turbulence in the settler and an inlet baffle as shown in Fig. 13 therefore is used.
The flow pattern in the settler is the next important consideration and has a significant effect on both the settler area required and the entrainment produced. Uneven flow in a rectangular settler can be caused by the settler inlet and outlet weirs not being full width. For this reason a “picket-fence” baffle is sometimes located immediately downstream of the inlet and immediately upstream of the outlet in order to distribute the flow as evenly as possible over the full settler width. A “picket-fence” baffle consists of two rows, one behind the other, of vertical bars placed across the width of the settler. The bars are placed so that those in the second row are located opposite open spaces in the first row. General Mills Inc. report that they have found the most effective spacing between the rows to be one bar width and that the actual dimensions of the bars appears relatively unimportant so long as the gap is approximately one-third to one-quarter the width of the bar. Not only is a “picket fence” useful in distributing the flow laterally across the settler, but it also effectively damps out wave motions which occur in the dispersion band, and which, if allowed to propagate, would cause dispersed phase to under or overflow the aqueous or organic weirs at the end of the settler.
The effect of the “picket-fence” as an aid to coalescence has been investigated using both organophilic and organophobic materials of construction and a small improvement has been observed when the fence is made of a material which is wetted by the dispersed phase. The use of plastic or wire mesh does under many circumstances assist coalescence and increases the allowable specific settler flow, although the susceptibility of the mesh to blockage makes its use less attractive on the large plant scale.
The settler area is determined by the specific flow, but this function gives no indication of the most effective shape for the settler. The criterion to be considered here is not the length/width ratio as is commonly thought, but is the linear velocity of the three phases in the settler and their relative velocities with respect to each other. The absolute velocity of the organic phase governs the crest height of the solution flowing over the organic weir and hence the minimum head loss achievable across the settler.
Too high a velocity in the organic layer also causes secondary haze and uncoalesced droplets to be entrained from the vicinity of the dispersion band and carried over the organic weir which greatly increases solvent losses.
The velocity in the dispersion band is governed by its depth and the rate of coalescence of the droplets within it. Linear flow within the band occurs towards locations where coalescence is occurring and consequently the dispersion band velocity decays exponentially until it becomes zero at the end of the settler. Backward flow is sometimes observed in parts of the settler, especially at the sides, but this can be minimized by the use of picket fences and a suitable inlet baffle.
Since linear velocity is a determining factor in settler design, the allowable length/width ratio of a settler decreases as the throughput is increased if the solution depths are maintained constant. The alternative is to increase the solution depths in the settler to lower the near velocity of each phase but this has the undesirable effect of increasing the organic inventory. For the LIX-copper systems which have been studied, solution depths of 1-ft 6-in. for the aqueous and 6-in. for the organic layers are considered minimum values if the specific settler flow is chosen to give a 4-in. dispersion band.
Solvent extraction process conclusion
The purpose of this paper has been to draw attention to significant factors in the design of mixer-settler units, particularly those for the recovery of copper using copper-specific reagents. Because of the high cost of reagent and the relatively low metal value price in these systems compared with other major solvent extraction applications, e.g., uranium recovery, optimization of the process with respect to solvent loss and extraction efficiency is particularly important. Therefore, factors which have not been of major importance in consideration of other systems have had to be explored in depth. In presenting some of the data produced in this exploration, the authors hope that other workers will be able to use the conclusions as a basis for work in different systems but would emphasize the great importance of evaluating each new, system on its own merits.