(8c) Prediction of Pressure Filtration Performance in Systems with Pharmaceutical High Aspect Ratio Crystals

Authors: 
Fragkopoulos, I. S., University of Leeds
Ahmed, B., University of Leeds
MacLeod, C., AstraZeneca
Muller, F. L., University of Leeds
Filtration is a key product recovery process in the pharmaceutical industry as it facilitates the separation of suspended solid particles, such as Active Pharmaceutical Ingredients (APIs) from a liquid. Filtration is an intermediate processing step between crystallisation and drying, and it typically involves several washing cycles aiding the removal of impurities and the manipulation of the mother liquor solubility [1]. Pressure filtration is the most common separation technique and its performance is strongly influenced by the cake structure and formation, which in turn are highly dependent on particle properties such as shape and size distribution. Cake filtration in systems with wide particle size distributions (PSDs) and asymmetrical high aspect ratio particles is frequently quite time consuming at scale. Rising the system’s pressure drop can accelerate the process, but it may also be counterbalanced by an increase in cake compressibility and/or particle breakage, both leading to denser packings [2]. The aim of this work is to study the cake filterability considering systems with high aspect ratio particles and various PSDs.

Lab scale filtrations have been conducted using a 200 ml, 35 mm diameter pocket filter using nitrogen pressure. Experiments have been conducted using dicalcium phosphate and mannitol. These materials were chosen because as pharmaceutical excipient the materials are commercially available with grades of different particle distributions. Filtration rates were compared for beds with different PSDs, at different applied pressures and for a settled bed or from a slurry. This approach was repeated using an API with high aspect ratio crystals that was wet milled to create three distinct PSDs and the filtration behaviour characterised in the same manner.

Modelling the filter cake structure is of crucial importance, since it allows for the accurate determination of cake properties. DigiDEM, a commercial Discrete Element Method (DEM) software, specially designed to deal with non-spherical shapes, with Lattice Boltzmann (LBM) built-in for drag force calculations [3], is used in this study to model the cake structure and consequently, evaluate the cake voidage as function of PSD and particle shape. DigiDEM is also used for the estimation of the particles’ terminal velocities due to gravitational sedimentation. With validated cake voidage and terminal velocities predictions at hand, the improved Ruth equation (an expansion of Darcy’s law) [4] is used in conjunction with lab-scale filtration data for the estimation of cake properties such as medium resistance, specific cake resistance and compressibility index. The estimated cake properties are then employed within the filtration model to predict the system response at higher pressures. Such a predictive model can be ultimately used for the successful system scale-up and control, and bring us a step closer to industrial adoption of a new Digital Design paradigm for pharmaceutical product and process design.

The authors gratefully acknowledge the support of the Advanced Manufacturing Supply Chain Initiative through the funding of the ‘Advanced Digital Design of Pharmaceutical Therapeutics’ project (Grant No. 14060).

References

  1. Rushton, A., A.S. Ward, and R.G. Holdich, Solid-liquid filtration and separation technology. 2nd ed. 2000: WILEY-VCH Verlag GmbH.
  2. Bourcier, D., J.P. Féraud, D. Colson, K. Mandrick, D. Ode, E. Brackx, and F. Puel, Influence of particle size and shape properties on cake resistance and compressibility during pressure filtration. Chemical Engineering Science, 2016. 144: 176-187.
  3. Structure Vision Ltd 2015, viewed 11 July 2017, http://www.structurevision.com/.
  4. Ruth, B.F., G.H. Montillon, and R.E. Montonna, Studies in filtration – I. Critical analysis of filtration theory. Industrial & Engineering Chemistry, 1933. 25: 76-82.