Maximising MIXING

Du Pont and Zeneca are among a number of multinationals seeking to increase the concentrations of reactants without compromising on reaction efficiency or product quality. They are funding research at the University of Birmingham, where Alvin Nienow's group is making headway in predicting optimum mixing conditions.

Waldemar Bujalski, Nienow's co-worker for the last 14 years, has been overseeing investigation into the suspension of particles at high concentrations. He leads a team of researchers from Japan, Italy and the Czech Republic.

'Pushing up the concentration of a solute makes the homogenisation problem more and more critical, especially if solids are associated with liquid phase reactions,' says Nienow.

Suspended solids have a significant effect on mixing time especially for concentrations greater than about 10 per cent by weight (w/w). The effect is greatest when solids are suspended such that they form a distinct layer at speeds (N) just below or above that needed to achieve suspension (Njs). Homogenisation is particularly slow for the clear liquid found above this slurry. Under these undesirable conditions, the time to achieve complete mixing (tm) can be increased enormously even sixfold.

It is well established that the addition of reactants to the top surface of a stirred vessel should be avoided, if possible. But the team's conclusions show that the method of adding a reactant is even more critical when slurries are being treated; its incorporation into the agitated mass may be extremely slow.

A stirred tank is arguably the most important type of reactor in production processes. Solids are incorporated into the continuous liquid phase and their successful suspension influences quality and quantity of the final product. For single phase systems, mixing times as a function of the power inputs and of the system's geometry are well known. But hitherto the influence of solids on the mixing time has been much less researched. Nienow believes that mixing time is strongly linked to solids concentration. 'Earlier studies did leave certain doubts concerning their results and because of their significance, further study seemed worthwhile', he says.

In general, under turbulent conditions, low power number impellers (axial flow hydrofoils pumping downwards, for example) have been found to be efficient because they can be used with a larger diameter and at low torque.

LIGHTNIN' THE LOAD

Nienow's team decided to investigate an axial hydrofoil impeller, the Lightnin A-315, in order to assess its potential usefulness at high solids concentration and to look at the influence of concentration on mixing time. This set-up would be relevant to many solid-liquid systems encountered in industrial practice, such as precipitation and crystallisation, hydration of lime, among others.

The researchers simulated industrial conditions, using water and sand. The sand was sieved such that particles had a mean diameter of ~900 m, of density 2500 kg/m3 at concentrations up to 20 per cent weight solids/weight liquid (w/w).

'In performing the research,' says Nienow, 'it is as important to observe general trends as absolute values because industrial situations are generally more complex than the model.'

The mixing time of the liquid under turbulent conditions was measured using a conductivity method developed for gas-liquid systems in which cages on the probes have been shown to allow a signal to be obtained without being disturbed by the presence of bubbles or solid particles. The use of tracers to monitor the location of mixing activity indicates that reactants should be fed into the impeller zone for more rapid incorporation in agitated slurries, rather than being dumped on top.

The solids concentration in the tank was increased at regular intervals from 0-20% w/w, and the minimum impeller speed required for complete suspension was found from visual observation of the particle motion. Suspension was 'complete' when no deposits stayed on the vessel base for more than 1-2s.

For the slurries, it was difficult to establish reliably the mixing time by decolourisation because of the colour of the sand and because of its attrition which was sufficient to cloud the water. Therefore, the technique was not effective for the slurry work.

Additional experiments using glass particles and ion exchange resins using Lightnin A-310 impeller and the decolourisation technique worked well. Similar trends were found and, at Njs for the low density small ion exchange beads, mixing times up to 1000 times that of water alone were found.

Nienow and Bujalski have identified three different mixing regimes :

(1) For speeds well below Njs, where very little solid was suspended, the measured mixing times were independent of solids concentration, very close to those of water and also similar for the three different probes positions.

(2) For speeds just below and up to Njs, but at which a significant movement of the solids at the vessel bottom was noticeable, the mixing times from mid height injection measured by the top probe were much longer compared with water. Those of the middle and bottom probes were noticeably so too. Under certain conditions the increase was up to a 1000 fold.

(3) For N significantly above Njs, the mixing time increases compared to water but much less dramatically.