Challenges for Scale-Up of Batch Phase Separation

Bi-phasic phase separations in batch processes often are operations which influence
cycle times and production capacity. A realistic assessment of phase separation time
for technical scale is important with respect to defining vessel sizes, process
step-time planning and an optimized adaption to further process steps. This requires
a rating of batch mixing & phase-separation at an early step of chemical-process development
in laboratory and a robust scale-up to the technical scale.
Currently available methods for design and optimization of such batch phaseseparations
show gaps in knowledge and in transfer from laboratory to technical
scale. Up to now it is not possible to trustfully predict mean drop diameter or dropsize
distribution for technical relevant bi-phasic systems in technical scale based on
laboratory tests. These are one of the key inputs to describe phase separation behavior,
i.e. sedimentation and coalescence. The prediction of drop-size distribution after
scale-up is important, since a remaining turbidity is significantly influenced by these.
In addition systems with higher viscosity show wider drop-size distributions and tendency
of turbidity will increase.
The specific industrial demand for a safe and validated scale-up method will be discussed.
Aspects are an integral general scale-up method for batch mixing and phase
separation, starting with standardized characterization of mixing and phaseseparation
behavior in lab and development of a scale-up method in order to evaluate
relevant aspects as equipment and mixing device, energy input, mixing time, phase
separation time, height of dispersion etc.
For modelling the performance of a batch settling accounting for polydisperse dropsize
spectra, the model of Henschke has been chosen as starting point, which accounts
for different average drop sizes. The Henschke model has been extended as
to include the sedimentation of drops with polydisperse drop-size distribution according
to the ReDrop approach (representative drops) by modelling an ensemble of individual
drops as they sediment and reach the close-packed layer. In principle coalescence
of the sedimenting droplets could be accounted for in this approach, but this
has been neglected in this first step.
The Henschke model already accounts for the time- and height-dependent average
drop sizes in the close-packed layer. This has been extended as to allow taking the
different drop sizes into account of those drops that are arriving at the close-packed
layer over time. The coalescence in the close-packed layer is described with the approach
used already by Henschke, which has been validated in principle by Kopriwa.
The results show that the polydispersity can well be described by this approach (see
Fig. 1). Especially – as is to be expected – the major phase separation occurs relatively
quickly leaving behind fine droplets in the continuous phase. Since this remaining
turbidity has a relatively low holdup, the sedimentation of the fine dispersion occurs
essentially with the sedimentation velocity of droplets in infinitely extended medium.
As a consequence the remaining turbidity as function of settling time can be
estimated with relatively good accuracy from the fraction of the fine droplets in the
original dispersion and their individual sedimentation velocity. It can also be seen that
the close-packed layer disappears after a certain settling time and after that the rate
of the arriving droplets is smaller than the time for their coalescence.
The open questions remaining will finally be addressed, namely the prediction of the
drop-size distribution after the mixing step, the detailed drop behavior within the close
packed layer as well as the influence of large-scale fluid dynamics within the settler.