Separating Vessels
4.3 Efficiency of Heavy Medium Separations
4.3.2 Factors Affecting the Separation efficiency
4.3.2.2 Effect of Viscosity and stability on Separation Efficiency
CHAPTER 4 SEPARATING VESSELS
By assuming that the main parameters of the separation are dependant upon the terminal velocity of the particle, the flow regime can be inferred by fitting values of a and b in an equation similar to Equation (4.1 0), with separating parameters such as P50 and Ep replacing Vt [Napier-Munn (1990)]. Most calculations, and direct estimates of terminal velocities, show that most particles separated in either static baths or dynamic separators move in the intermediate flow regime.
Larger particles in both types of separators will experience fully turbulent flow, in which viscosity plays no role, whereas the finest near-gravity particles separated in dynamic separators like cyclones will approach laminar flow, in which the influence of medium viscosity is very high.
The problem with heavy medium suspensions is that at high medium specific gravities and high solid volume concentrations, their behaviour becomes non-Newtonian. In static baths another problem is that hindered settling sometimes prevails over free settling conditions. In dynamic separators, due to the high centrifugal forces, the particles comprising the heavy medium also undergo some classification. These factors compromise the separation efficiency of heavy medium separators, and make the prediction of partition curves and separation efficiency difficult.
CHAPTER 4 SEPARATING VESSELS
dynamic separators. Most heavy medium separations take place in the intermediate flow regime, in which the influence of viscosity on terminal velocity, and thus on Pso and Ep, is variable.
Napier-Munn (1990) discusses the results of other workers on the separation of coal in bath separators. The coal separated had a top size of 50 mm, and the lower size varied from 3 to 6 mm.
in each case, the efficiency of separation declined as viscosity increased. This corresponds to an increase in Ep values with an increase in viscosity. It was also observed that the viscosity effects were strongest for coals with a high proportion of near-gravity material. In a separate investigation, efficiency data were correlated as follows:
( J
1/2
Ep =0.0085
~
(4.11)where IIp == plastic viscosity (cP), and d == feed size (inches).
By comparing the exponents in Equation (4.11) to those in Equation (4.10), it was concluded that the particles moved mainly in the turbulent or upper intermediate regime. As shown in Section 4.3.2.1, this means that viscosity has only a little effect, if any, on the separations occurring in static baths. Experiments on Drewboy separators confirmed that turbulence prevailed in these separators, and that the main problem faced in bath separators is that coarse settling particles have a tendency to drag small particles with them.
In the early development of dynamic separators, it was generally believed that viscosity had little influence on their separation efficiency, because of the high rates of shear prevailing [Napier- Munn (1990)]. However, numerous experimental works have proven that viscosity does playa role, although the effects may be small at low medium densities and low contamination levels.
Collins et al. (1983) showed that both suspension viscosity and stability have an effect on the separation efficiency of dynamic separators. They showed that a suspensions with a high viscosity resulted in lower separation efficiencies. They also showed that although lower viscosities resulted in higher efficiencies, there was a critical value beyond which the low stability of the suspension also resulted in inefficient separations. They showed that the separation efficiency of fme particles was much higher at lower suspension viscosities.
CHAPTER 4 SEPARATING VESSELS
In the analysis of heavy medium separations using dynamic separators, particularly cyclones, other parameters apart from Ep values have been used to infer separation efficiency. One such parameter is the density of separation in the cyclone, which represents the density at which 50 % of the feed has an equal chance of reporting to either the overflow or under flow. The separation density is directly dependant on the amount of material reporting to the floats and sinks product.
A lower separation density results in higher proportions of ore to the underflow, whereas a higher separating density results in a higher proportion of ore to the overflow. The separation density is not necessarily equal to the medium density, and this is confirmed by the following expression [Napier-Munn (1990)):
(4.12)
where Dc
=
cyclone diameter; Q=
flowrate; K=
geometry constant; Q=
feed medium density, and /l = apparent viscosity.Using dimensional analysis and carrying out experiments to confirm the analysis, Napier-Munn (1990) concluded that for stable media, both the separation density and the Ep increase with viscosity. This can easily be deduced from Equations (4.11) and (4.12).
Collins et al. (1983) showed that the separation density and Ep are a quantitative function of the medium density, normally the underflow density. They performed experiments on a Vorsyl separator and DSM cyclone using ferrosilicon and ferrosilicon / magnetite suspensions. They found that the viscosity and stability had a direct impact on the separating density and Ep of the separators. To infer the amount of particle segregation occurring in the separators, due to the unstable nature of heavy medium suspension, the density differential was used (Q - Q), where
C1
and Q are the underflow and overflow densities, respectively. The segregation of the medium also has an effect on the relationship between the underflow density and the separation density.
For relatively low values of density differential, the separation density is usually higher than the underflow density. However, as these values increase, the separating density begins to drop below the underflow density. Under conditions of medium stability the separation density is equated to the underflow density for all types of media for the Vorsyl separator DSM cyclone used in the latter's work.
CHAPTER 4 SEPARATING VESSELS
Analysis of the product streams from a Vorsyl separator indicated that little segregation of the medium occurred by particle size or density with cyclone differentials ([1- (1) up to 500 kgm-3_
Generally, segregation only became a serious problem when the differential exceeded 700 kgm-3 for a 1 / 1 cyclone 60 (and coarse FeSi) - magnetite mixture. It was then established that for this type of medium the cyclone underflow density increased linearly with cyclone feed density up to a differential of 500 - 600 kgm-3 and then stabilized at a constant density, which indicates that classification of the medium was occurring at this point.
Testwork conducted on the DSM cyclone also showed the same results as those for the Vorsyl separator. In this case, however, the transition point occurred at 800 kgm-3. It was also found that although the separation density equated to the cyclone underflow density below this differential, it dropped rapidly above this level.
In order to maximise the efficiency of separations in both the DSM cyclone and the Vorsyl separator, an optimum density differential has to be reached. This value is also highly dependent on the ore particle size. Generally, the best working range for both separators is with a density differential of200 - 500 kgm-3, the optimum being at about 400 kgm-3 [Collins et al. (1983)].