El TRIBUTO

Oosterhuis (1985) summarised how regime analysis and scale down can be used to solve scale up problems in the following scheme:

PRODUCTION SCALE 1. Regime analysis Simulation Application Optimisation LABORATORY SCALE

In regime analysis, the first stage is to ascertain which is the ruling regime in a particular operation. There may be one (pure) or two (mixed) regimes, or even a change of regime during a time-related process or on scale-up. It is important to define a characteristic parameter on which to base laboratory scale experiments, especially before the construction of any specialised equipment.

Geometric similarity o f the equipment simulating the real operation may not be required if the ruling regime is better modelled with an alternative geometry: the important criteria at this stage is that laboratory experiments must be representative of what goes on at production scale.

Optimisation of the process at laboratory scale must be translatable to production scale: any extrapolation of empirical or mechanistic models will be limited by practical constraints of large scale operation.

Sweere et al (1987) proposed an extension to the scheme where the first step in the scale down procedure not only includes regime analysis, but also

dimensional analysis, mechanistic analysis and the similarity principle. Complementary knowledge about the process was also supplied from rules of thumb, literature data, correlations and experience.

If all these considerations have been taken into account, then scale up is based upon the application of a reversal of the methods used to scale down. Further examples where scale down techniques have been applied are: gas-liquid flow in pipes (Chesters 1977), trickle flow reactors (Gierman 1988), chromatography (Naveh 1991, Hinrichsen 1985), testing of automated production software (Young et al 1984), and disc stack centrifuges (Mannweiler and Hoare 1992).

1.3 SCALE DOWN OF CENTRIFUGE OPERATIONS

There are several types of industrial centrifuge, which are reviewed in standard texts such as Perry (1986), Coulson and Richardson (1985), Svarovsky (1990), and the main types are listed in Table 1.3.1. Centrifuges are frequently selected on the basis of their solids handling capacity and ability to recover small particles as shown in Table 1.3.1 and Figure 1.3.1 Centrifuge modifications for biotechnology include bowl cooling, clean in place and steam sterilisation.

Table 1.3.1: Types o f industrial centrifuge and their selection with respect to solids handling.

Centrifugal separation equipment solids handling capacity (% v/v)* Solids discharge type Tubular < 5 manual Multichamber < 5 manual

Scroll decanter 15-60 continuous

Disc stack Solids ejecting 0 - 1 0 intermittent

Nozzle 10-30 continuous

Figure 1.3.1: Performance o f various centrifugal sedimentation equipment. 100 / I S c r o l l B a s k e t t y p e Di s c Tu b u l a r C L a b o r a t o r y d i s c _{G r a v i t y t a n k} L a b o r a t o r y t u b u l a r 0 01 0 001 5 10 2 0 5 0 0 0 1 0 0 2 0 0 5 0 1 0 2 0 5 1.0 2 P a r t i c l e s i z e x , p m , d e n s i t y 2 0 0 0 kg m' ^ in w a t e r 10^ io-^ Q / Z = 2 X s e t t l i n g v e l o c i t y u n d e r gr a v i t y , m S"'

The centrifuges examined in this study can be operated with continuous throughput. The intermittent solids discharge disc stack centrifuge is a high capacity clarifier, which is able to recover small particles. The scroll decanter centrifuge can handle high feed solids concentrations and discharges highly dew atered solids. A n understanding o f the param eters w hich are likely to govern their scale dow n may be found from established scale up procedures and flow path inform ation found in the relevant literature.

1.3.1 Scale up or down using sigma theory

The sigma value concept first proposed by A m bler (1952) is used for predicting separation characteristics in industrial centrifuges.

In general:

e = s

W here Q = volumetric flow rate

Ug = terminal velocity o f a solid particle settling under gravity through a liquid

E = sigma factor

Term inal settling velocity (Ug) is found from Stoke's law for a free particle m oving through a liquid under ideal settling conditions (section 3). E is the

(m's'^)

(m s '')

factor by which centrifugal acceleration increases settling capacity relative to the area of a gravity settling tank.

When scaling up or down with the sigma value, it is assumed that Ug remains constant in geometrically similar centrifuges, and that separation is a function of flow rate capacity and equivalent settling area alone:

Ql = Q l 4

E;

Large discrepancies between prediction and practice using this method are usually attributed to deviations from Stoke's law due to changes in the degree of turbulence, or a change in the break-up of aggregated particles due to shear.

Efficiency factors reduce E on scale up, and vary in magnitude according to the extent of departure from idealised sedimentation in each type of centrifuge.

The efficiency factor for scaling up bottle centrifuges used for laboratory analysis is 90-100% (Perry 1986). However, in scale up of industrial centrifuges such as tubular bowls, there is a deterioration in performance with an efficiency factor of 80% (Perry 1986). For disc stack centrifuges, efficiency factors vary from 73% for a dilute monodisperse suspension (Murkes and Carlsson 1978) to 55% (Ambler 1959/1961 and Frampton 1963), 45% (Morris 1966) and 40% (Purchas 1981). The reduction in effective clarification area on scale up of helical conveyors varies from 54%-67% (Ambler 1959, Morris 1966, Sokolov 1971) depending on the volume occupied by sediment. However, in this case scale up is also limited by the ability to convey solids. The magnitude of the efficiency factor becomes more pronounced in larger centrifuges. The problem faced in scale down to a small centrifuge is how to mimic the performance of a large machine without being unduly optimistic.

Disc stack centrifuges have also been scaled up using "KQ" value which is inversely proportional to Ug (Sullivan and Erikson 1961). K is found from empirical measurements on a bowl with known geometry and particle-related constants.