## Abstract

Crystallization is an important separation process to obtain high value-added chemicals in crystalline form from liquid solution in pharmaceutical, food and fine chemical industries. As in most of the particulate processes, the quality of the solid product is determined by its particle size distribution (PSD), which is the result of the combination of events at the microscopic and macroscopic scale. The microscopic events are governed by complex kinetic interactions between the solute molecules and the crystal lattice, and diffusion mechanisms, whereas the macroscopic events are related to the crystallization operation (solute concentration, temperature profile, mixing).

Traditionally, the crystallization has been operated in batch or semi-batch mode. However, batch operations suffer from disadvantages such as lack of batch to batch reproducibility, long processing times, scale up issues, poor controllability and observability (Porru and Ozkan, 2016). Hence recent research efforts are directed towards the design of continuous crystallization technologies, which may be able to overcome the above mentioned limits of the batch operation. A number of continuous crystallizer types and configuration have been applied, such as the mixed suspension, mixed product removal (MSMPR) crystallizers in single or multiple stage configurations (Alvarez et al., 2011), Plug Flow Crystallizers (PFC), with (Alvarez and Myerson, 2010) or without static mixers, and more recently, PFC with recycle (Cogoni et al, 2015).

This work addresses the optimal design of a continuous flash cooling crystallization system. This technology enhances the crystallization phenomena through the evaporation of the solvent at low temperatures by operating under vacuum and vapour-liquid equilibria. In particular, we are interested in the identification of a suitable configuration of single or multiple MSMPR crystallizers with or without recycle in order to achieve certain process targets, namely production yield and relevant attributes of the PSD. More specifically, we are interested to obtain a process configuration amenable to achieve good compromise between yield and large average dimension size of crystals by optimizing the recycle ratio and pressure of the vessel. To this end, an optimization problem which maximizes the product yield is formulated for the configurations

i) a single MSMPR crystallizer

ii) a single MSMPR crystallizer with recycle

iii) a series of two MSMPRs, without intermediate separation of the solid product and without recycle.

The constraints for the above mentioned optimization problem are derived from the definition of production targets, namely relevant PSD attributes of the final product (in terms of 50th percentile, D50), daily production, and upstream operation (temperature and flow rate of the inlet flow). The optimal design study is performed using a crystallization process model with the kinetics of size independent growth and secondary nucleation due to crystal impeller collision

The performance of the three configurations are studied through simulations at different pressures and recycle ratios, which were necessary to cast a feasible optimization problem. From this simulation study it has been found that (i) the solute concentration is about at saturation, which depends on the crystallization kinetics used. For the cases C1 and C2 the decreasing of the pressure leads to (ii) an increase of the yield and (iii) a decrease of the D50. (iv) The introduction of a recycle stream is not improving the yield nor the dimension of the crystals. The analysis of the case C3 leads to that (v) the pressure in the first crystallizer (P1) must be higher than the one in the second crystallizer (P2) in order to avoid dissolution, and (vi) the yield only depends on P2 while (vii) the D50 of the final product depends on the combination of P1 and P2.

The optimal design study has shown that in the first configuration, the yield is maximized at the minimum allowed pressure (and thus, minimum allowed temperature). If a quality constrain for the D50 is set, then the optimal yield is obtained at a higher pressure which actively satisfy the above mentioned constrain. In the second configuration, it is not possible to obtain an optimal solution since the recycle ratio has no influence on the yield and the D50. In case of the third configuration, it has been found that at a fixed P2 (and thus at fixed yield), the D50 of the final product has a quadratic functionality with P1. Thus, the optimization problem can be reformulated in order to find P1 which maximize the D50, under the constrain of the minimum allowed P2. This P2 is the optimum of the unconstrained optimization problem casted for the single crystallizer (first configuration). When the single MSMPR and the series of MSMPRs operate at the same yield, the third configuration allows to obtain larger crystals than the first configuration.

The optimal design of three different configurations of MSMPRs for flash cooling crystallization has led to the conclusion that the optimal pressure for the single MSMPR is the result of a compromise between high yield and large D50. Although the addition of a recycle stream does not improve the crystallization performance, the addition of a second crystallization unit does. Indeed an optimal setting of the pressure of the first and second crystallizer allows to operate at the maximum yield and obtaining larger crystal compared with the use of a single unit.

An evaluation of the annual expenses associated with the crystallization is under consideration in order to evaluate if the cost of adding a second unit is paid back by the improvements in the dimension of the crystals.

Acknowledgments

This work has been done within the project Improved Process Operation via Rigorous Simulation Models (IMPROVISE) in the Institute for Sustainable Process Technology (ISPT).

References

Alvarez, A.J., Myerson, A.S., 2010. Continuous plug flow crystallization of pharmaceutical compounds, Cryst. Growth Des. 10 (5), 2219–2228.

Alvarez, A.J., Singh, A., Myerson, A.S., 2011. Crystallization of cyclosporine in a multistage continuous MSMPR crystallizer, Cryst. Growth Des. 11 (10), 4392–4400.

Cogoni G., de Souza B.P., Frawley P.J. 2015 Particle size distribution and yield control in continuous crystallizers with recycle, Chemical Engineering Science 138, 592-599.

Porru, M. and Ozkan, L. Systematic observability and detectability analysis of industrial batch crystallizers, accepted paper at DYCOPS 2016.

Traditionally, the crystallization has been operated in batch or semi-batch mode. However, batch operations suffer from disadvantages such as lack of batch to batch reproducibility, long processing times, scale up issues, poor controllability and observability (Porru and Ozkan, 2016). Hence recent research efforts are directed towards the design of continuous crystallization technologies, which may be able to overcome the above mentioned limits of the batch operation. A number of continuous crystallizer types and configuration have been applied, such as the mixed suspension, mixed product removal (MSMPR) crystallizers in single or multiple stage configurations (Alvarez et al., 2011), Plug Flow Crystallizers (PFC), with (Alvarez and Myerson, 2010) or without static mixers, and more recently, PFC with recycle (Cogoni et al, 2015).

This work addresses the optimal design of a continuous flash cooling crystallization system. This technology enhances the crystallization phenomena through the evaporation of the solvent at low temperatures by operating under vacuum and vapour-liquid equilibria. In particular, we are interested in the identification of a suitable configuration of single or multiple MSMPR crystallizers with or without recycle in order to achieve certain process targets, namely production yield and relevant attributes of the PSD. More specifically, we are interested to obtain a process configuration amenable to achieve good compromise between yield and large average dimension size of crystals by optimizing the recycle ratio and pressure of the vessel. To this end, an optimization problem which maximizes the product yield is formulated for the configurations

i) a single MSMPR crystallizer

ii) a single MSMPR crystallizer with recycle

iii) a series of two MSMPRs, without intermediate separation of the solid product and without recycle.

The constraints for the above mentioned optimization problem are derived from the definition of production targets, namely relevant PSD attributes of the final product (in terms of 50th percentile, D50), daily production, and upstream operation (temperature and flow rate of the inlet flow). The optimal design study is performed using a crystallization process model with the kinetics of size independent growth and secondary nucleation due to crystal impeller collision

The performance of the three configurations are studied through simulations at different pressures and recycle ratios, which were necessary to cast a feasible optimization problem. From this simulation study it has been found that (i) the solute concentration is about at saturation, which depends on the crystallization kinetics used. For the cases C1 and C2 the decreasing of the pressure leads to (ii) an increase of the yield and (iii) a decrease of the D50. (iv) The introduction of a recycle stream is not improving the yield nor the dimension of the crystals. The analysis of the case C3 leads to that (v) the pressure in the first crystallizer (P1) must be higher than the one in the second crystallizer (P2) in order to avoid dissolution, and (vi) the yield only depends on P2 while (vii) the D50 of the final product depends on the combination of P1 and P2.

The optimal design study has shown that in the first configuration, the yield is maximized at the minimum allowed pressure (and thus, minimum allowed temperature). If a quality constrain for the D50 is set, then the optimal yield is obtained at a higher pressure which actively satisfy the above mentioned constrain. In the second configuration, it is not possible to obtain an optimal solution since the recycle ratio has no influence on the yield and the D50. In case of the third configuration, it has been found that at a fixed P2 (and thus at fixed yield), the D50 of the final product has a quadratic functionality with P1. Thus, the optimization problem can be reformulated in order to find P1 which maximize the D50, under the constrain of the minimum allowed P2. This P2 is the optimum of the unconstrained optimization problem casted for the single crystallizer (first configuration). When the single MSMPR and the series of MSMPRs operate at the same yield, the third configuration allows to obtain larger crystals than the first configuration.

The optimal design of three different configurations of MSMPRs for flash cooling crystallization has led to the conclusion that the optimal pressure for the single MSMPR is the result of a compromise between high yield and large D50. Although the addition of a recycle stream does not improve the crystallization performance, the addition of a second crystallization unit does. Indeed an optimal setting of the pressure of the first and second crystallizer allows to operate at the maximum yield and obtaining larger crystal compared with the use of a single unit.

An evaluation of the annual expenses associated with the crystallization is under consideration in order to evaluate if the cost of adding a second unit is paid back by the improvements in the dimension of the crystals.

Acknowledgments

This work has been done within the project Improved Process Operation via Rigorous Simulation Models (IMPROVISE) in the Institute for Sustainable Process Technology (ISPT).

References

Alvarez, A.J., Myerson, A.S., 2010. Continuous plug flow crystallization of pharmaceutical compounds, Cryst. Growth Des. 10 (5), 2219–2228.

Alvarez, A.J., Singh, A., Myerson, A.S., 2011. Crystallization of cyclosporine in a multistage continuous MSMPR crystallizer, Cryst. Growth Des. 11 (10), 4392–4400.

Cogoni G., de Souza B.P., Frawley P.J. 2015 Particle size distribution and yield control in continuous crystallizers with recycle, Chemical Engineering Science 138, 592-599.

Porru, M. and Ozkan, L. Systematic observability and detectability analysis of industrial batch crystallizers, accepted paper at DYCOPS 2016.

Original language | English |
---|---|

Publication status | Published - 13 Nov 2016 |

Event | 2016 AIChE Annual Meeting - San Francisco, United States Duration: 13 Nov 2016 → 18 Nov 2016 http://www.aiche.org/conferences/aiche-annual-meeting/2016 |

### Conference

Conference | 2016 AIChE Annual Meeting |
---|---|

Country | United States |

City | San Francisco |

Period | 13/11/16 → 18/11/16 |

Internet address |