Modeling and control of crystal shape in continuous protein crystallization

Crystal shape and size control using a plug flow crystallization configuration

“…Motivated by this, a mixed suspension mixed product removal (MSMPR) crystallization process, which is analogous to the conventional continuously stirred tank crystallizer (CSTC), has received a growing attention, and many efforts have been made in order to produce crystals from the MSMPR process with a higher production rate and desired product quality (Griffin et al, 2010;Alvarez et al, 2011;Hou et al, 2014;Ferguson et al, 2014). However, due to the presence of back-mixing, which is commonly modeled by employing the residence time mixing model, those crystals nucleated at a later stage during the crystallization process will reside a relatively short amount of time in the crystallizer and thus they will end up leaving the crystallizer with undesired size and shape distributions (Kwon et al, 2014). To this end, plug flow crystallizer (PFC) has been proposed to produce crystals with narrow size and shape distributions (Eder et al, 2011;Vetter et al, 2014).…”

“…Therefore, the necessity of incorporating the constraints to account for the physical limitations on the manipulated inputs and operating conditions makes the model-based control strategy (Miller and Rawlings, 1994;Worlitschek and Mazzotti, 2004;Shi et al, 2005;Mesbah et al, 2010) the method of choice for crystal size distribution control. Specifically, the model predictive control (MPC) scheme was employed by Kalbasenka et al (2007) and Kwon et al (2013Kwon et al ( , 2014 in order to control the crystal size and shape distributions along with the consideration of the crystal growth and nucleation processes in both batch and MSMPR processes based on a reduced-order model. Furthermore, in addition to model-based optimization to compute optimal jacket temperature values, a feed-forward control (FFC) is proposed in the present work for the production of crystals with desired size and shape distributions owing to its unique ability to deal with feed flow disturbances that occur during the operation of the PFC though the use of the online measurements of the inflow solute concentration, PFC temperature, and crystal seed size.…”

Measurement, modelling, and closed-loop control of crystal shape distribution: Literature review and future perspectives

“…Therefore, the effective dimensionality of the problem became very small, allowing simulation of a population of asymmetric crystals with minimal computational effort. Other research in multi-dimensional and morphological PB modelling of crystallisation processes include the work of Majumder and Nagy [102] who studied the influence of crystal growth modifiers, Kwon et al [40,103] and Liu et al [104][105][106] who studied the simulation, optimisation and control of protein crystals.…”

“…The genetic algorithm was used to perform the optimization. Christofides and co-workers [103] also investigated the modelling and control of crystal shape in continuous protein crystallisation using again the aspect ratio of mean sizes between faces (101) and (110) as the objective function. The formed objective formulation includes the weighted aspect ratio term, growth ratio term and jacket temperature change term.…”

“…When using a target shape as the objective function, the mean sizes of individual faces from their size distributions or the ratios between the mean sizes at each optimizing time are normally used as the optimization variables as described in literature [103,104,111]. If the shapes of size distribution for individual faces are conserved during the crystallisation processes, for example the crystal growth from solution is dominated by pure growth mechanisms with little effects from size itself, secondary nucleation, aggregation and breakage, the standard deviation for each size distribution of a crystal face will be kept the same over the whole crystallisation process.…”