Professional Documents
Culture Documents
net/publication/320150546
CITATIONS READS
7 287
9 authors, including:
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by Humera Siddique on 25 March 2019.
a r t i c l e i n f o a b s t r a c t
Keywords: This paper presents the application and challenges of achieving Model Predictive Control (MPC) on two
Continuous crystallisation continuous oscillatory baffled crystallisation reactors, delivering precise product quality control in the face of
Super-saturation control raw material fluctuations. A key advantage of MPC is that it effectively deals with multivariable interactions
Model predictive control and constraints that appear within the continuous crystallisation process. Using a flexible real-time software
Calibration models
package, a control scheme is proposed that incorporates three MPC blocks for controlling reactor cooling profile,
Crystal PSD
API (Active Pharmaceutical Ingredient) concentration and crystal size distribution, respectively. Furthermore,
the solution is customisable and transferable to different crystallisation reactors as well as various APIs.
© 2017 Elsevier Ltd. All rights reserved.
1. Introduction to continuous crystallisation closed loop super-saturation control is seldom used during routine man-
ufacturing. This often results in inconsistent particle size distribution
Crystallisation is a critical production step in the Pharmaceutical from one batch to another. Moreover, batches containing higher levels
Industry. Whether it is for purification of the final Active Pharmaceutical of fines severely limit the effectiveness of downstream processes such as
Ingredient (API) or synthesis of compounds, it is necessary to maintain filtration and drying.
precise process conditions to achieve a product which meets quality These problems have led the Pharmaceutical Industry to seek contin-
specifications. uous crystallisation methods as an alternative for batch manufacturing.
Batch processing, mainly using large continuously stirred tank re- The aim is to discard the need for reactor scale up, which would mean
actors (CSTRs), is the most common approach to crystallisation for that the process designed in the laboratory could be readily reproduced
commercial manufacturing. The crystallisation process for a given on a manufacturing plant (Zhao et al., 2014). Moving the crystallisation
compound is typically developed at a small scale in a research labo-
process from batch to continuous would also mean better concentration
ratory. For simple cool-down crystallisation, PAT (Process Analytical
control, more consistent product and hence lower manufacturing costs.
Technology) enabled experimental techniques can be used to establish
Tighter control of the processes would also facilitate implementation of
the compound’s Meta-Stable Zone (MSZ) and super-saturation control
the Quality-by-Design (QbD) techniques. Attaining more precise control
strategies (Fig. 1). During these experiments the mixing rates, seed
of product attributes using a combination of PAT and Advanced Control
loading and cooling profiles are optimised to achieve desired properties
techniques fulfils the QBD objectives; doing so for a continuous process
of crystals, such as their morphology or particle size distribution. Once
these conditions are established, the process can then be scaled up to enhances the versatility and also addresses the need for increased
manufacturing volumes. manufacturing flexibility.
As the process is scaled up, accurate control of the reactor becomes Continuous Oscillatory Baffled Reactors (COBRs) provide an alterna-
more challenging. The Meta-Stable Zone is not only compound specific, tive to CSTRs, where mixing is achieved through oscillatory movement
but also reactor specific. Process parameters such as impeller design, of the fluid though narrower parts of the reactor. Here, a time dependent
mixing shear force and heat-transfer characteristics can shift the meta- cooling profile is transformed into a length dependent one, where
stable zone significantly from the small-scale parameters. In addition, residence time can be adjusted by varying flow of feed solution. A single
* Corresponding author.
E-mail addresses: ftahir@perceptiveapc.com, ftahir49@hotmail.com (F. Tahir).
1 Co-first author.
http://dx.doi.org/10.1016/j.conengprac.2017.07.008
Received 1 November 2016; Received in revised form 18 July 2017; Accepted 21 July 2017
0967-0661/© 2017 Elsevier Ltd. All rights reserved.
F. Tahir et al. Control Engineering Practice 67 (2017) 64–75
reactor of up to several metres in length can be assembled by combining 1.1. Rattlesnake crystallisation reactor
multiple straights. Since it is possible to achieve plug flow through
the reactor, a COBR is a natural choice for continuous crystallisation. Rattlesnake is a COBR unit developed by Cambridge Reactor De-
Moreover, a COBR allows for uniform and efficient mixing, rapid heat sign (Cambridge Reactor Design, 0000). A lactose crystallisation study,
transfer and reduced sheer. Literature describes several crystallisation reported in (Siddique et al., submitted for publication), showed that
transfers from batch OBRs to COBRs and advantages of continuous this continuous crystallisation unit is able to deliver 26% higher yield
crystallisation (McGlone et al., 2015; Siddique, Brown, Houson, & and significant reduction in particle span as compared to its batch
Florence, 2015). counterpart.
A number of control/monitoring schemes have been proposed in the The unit consists of four cylindrical modules, as shown in Fig. 2.
literature for the crystallisation process. In Orkoulas, Kwon, Nayhouse, Each module has a length of 740 mm and an internal diameter of 69 mm
and Christofides (2014), the authors propose a comprehensive dynamic (giving an approximate volume of 2760 ml). To control the temperature,
model for a continuous plug flow crystallisation unit based on the Mass, each module is fitted with a double water filled jacket: a primary
Energy and Population balances. This model is then used to compute a jacket with co-current configuration and a secondary coil with counter-
set of optimal jacket temperatures for the crystalliser as well as flow ve- current flow. As shown in Fig. 2, these are connected to the hot and
cold circulators, respectively. Such a design enables precise temperature
locity which minimises the crystal shape and size deviations from their
control which is important for maintaining an accurate cooling profile
desired values. The scheme also takes account of disturbances to the
during the crystallisation. The residence time in the reactor is generally
inflow solute concentration by computing new jacket temperature. In Li,
a function of the throughput and varies from approximately 60 min
Mhaskar, Shi, El-Farra, and Christofides (2006), the authors present a
upwards. Further information about the unit is available in Cambridge
population balance model as well as a hybrid predictive control scheme Reactor Design (0000).
to stabilise the oscillatory crystallisation dynamics (that occur due to With regards to the experimental setup, product is fed through
complex interplay of nucleation and growth) as well as control particle two peristaltic pumps—for feed solution and seed slurry, respectively.
size. In Chiu and Christofides (1999), a population model and nonlinear Additionally, oscillatory feed-flow is achieved by a push–pull hydraulic
control scheme is developed to minimise the oscillations associated with oscillator, as indicated in Fig. 2. Different product properties in the reac-
a homogeneous (CSTR-type) continuous crystallisation unit, obtain a tor (e.g. temperature, and particle size) are monitored (and controlled)
desired PSD as well as guarantee closed-loop stability. Kwon, Nayhouse, through a set of thermocouples and PAT devices. In particular, we use
Christofides, and Orkoulas (2014) used both population balance mod- Focused Beam Reflectance Measurement (FBRM) for particle size and
elling and MPC to control shape and size. The model, with one actuator Fourier Transform Infrared (FTIR) spectrometer to predict the real-time
(jacket temperature) was simulated and showed that MPC can reduce solute concentration (through the development of a calibration model—
the impact of disturbances on the crystallisation process. This approach see the sections below).
was developed further by Mack (2013), where MPC was applied to the
temperature cooling profile, while ensuring that desired concentration 1.2. DN15 crystallisation reactor
was achieved. Furthermore, Forgione et al. (2015) applied, on a test
plant, an iterative algorithm which adjusted the cooling profile at The DN15 is another type of COBR developed by Nitech and
CMAC (Nitech Solutions, 0000). As shown in Fig. 3, it consists of
each batch, while the product temperature was controlled with a PI
31 straights (tubes), each containing of a series of multiple baffle
controller.
and inter-baffle zones. The reactor internal diameter is 15 mm and
In this work, a novel advanced control scheme is proposed for the
a length of 14 m giving a total volume of approximately 2500 ml.
plug flow continuous crystallisation process. The main objectives of this
The tubes are jacketed, providing liquid flow in order to control the
solution are to control the reactor cooling profile, API concentration product temperatures throughout the reactor. The temperature range
and crystal particle size. Due to plug flow dynamics, some of the is from 0 to 100 ◦ C and the reactor is fitted with 39 thermocouples
aforementioned issues around oscillations are avoided. Furthermore, and 13 circulators to maintain the temperature profile. PAT equipment,
due to real-time feedback from the inline PAT, the proposed control including FTIR spectrometer and FBRM, is integrated into the reactor
scheme is able to deal with various disturbances such as reactor fouling at certain positions. Similar to the Rattlesnake unit, DN15 is also able
and feed impurity. The scheme considers Lactose as the API and is tested to achieve plug-flow with residence times from approximately 60 min
on two commercially available continuous reactors: Cambridge Reactor upwards (depending on the flow). A schematic of the DN15 is shown in
Design’s Rattlesnake COBR and Nitech’s DN15 COBR (see Sections 1.1 Fig. 3. Further information on the unit is available at (Nitech Solutions,
and 1.2, respectively). 0000).
65
F. Tahir et al. Control Engineering Practice 67 (2017) 64–75
2. Model structure for COBR dynamics where 𝐲𝐤 is an 𝑛-dimensional vector of Controlled Variables (CVs),
and 𝐮𝐤 is an 𝑛𝑢 -dimensional vector of Manipulated Variables (MVs),
all at sample time 𝑘. The symbol 𝑠𝑢 denotes the delay spread for the
Data driven models for the COBR unit have been identified using MVs, and 𝐀𝟏 , 𝐁𝐢 , ∀𝑖, are appropriately-dimensioned matrices of model-
the RLS (Recursive Least Square) algorithm. This has the advantage of coefficients. Finally, the 𝚫 in Eq. (1) denotes incremental signals such
relatively fast calculations and good convergence properties (Goodwin that:
& Sin, 2009). In particular, three Output Error (OE) models are identified
𝚫𝐮𝐤 = 𝐮𝐤 − 𝐮𝐤−𝟏 .
for predicting cooling temperature, concentration of Lactose and crystal
size distribution, respectively (see Remark 1). The structure of the OE Note that the considered model (1) has an incremental structure (as
prediction model considered is of the form: opposed to absolute). Such a model format has the advantage of
robust identification of model-coefficients as well as rejection of any
𝑠𝑢
∑ unmeasured drift within the process (see e.g. Maciejowski, 2002 and
𝐲𝐤+𝟏 − 𝐲𝐤 = 𝐀𝟏 (𝐲𝐤−𝟏 − 𝐲𝐤 ) + 𝐁𝐢 𝚫𝐮𝐤−𝐢 (1) the references therein).
𝑖=0
66
F. Tahir et al. Control Engineering Practice 67 (2017) 64–75
67
F. Tahir et al. Control Engineering Practice 67 (2017) 64–75
68
F. Tahir et al. Control Engineering Practice 67 (2017) 64–75
example, the dead time (pure delay) for the ’seed flow to Concentration’ Therefore, the focus of this paper is to develop a flexible solution that
loop is approximately 1 h with settling time of around 1.5 h. All this can be transferred between products and reactors thus reducing time
serves to increase the size of the optimisation problem. Due to this and effort. Fig. 5 shows the proposed approach to the development of a
reason, horizon compression is used within the MPC scheme to ensure transferable APC solution for continuous crystallisation. It is designed to
that the optimisation problem is solved in a timely manner with regards be fully adaptable to different crystallisation reactors with temperature-
to the control interval (see Section 4 for details). based concentration control.
The proposed APC solution is implemented in PharmaMV real-
4. Transferable APC solution time system (Perceptive Engineering Limited, 0000) and consists of
three Model Predictive Controllers (with two of them in a cascade
As stated above, MPC is particularly well suited to continuous arrangement—Fig. 5). Since crystallisation has relatively slow dynam-
crystallisation reactors as they often exhibit multivariable interactions, ics, the entire process and PAT data is sampled every 5 s. The first MPC
constraints, significant time delays and slow process responses. The provides control of the reactor temperature profile and has an interval
MPC employs dynamic process model predictions and feedback to of 60 s for both DN15 and Rattlesnake reactor (further details are given
compensate for both measured and un-measured process disturbances. in Section 5). The particle size MPC controller, with an interval of 480 s,
However, development of such control systems can be time consuming. uses FBRM measurements in real-time. This controller achieves the
69
F. Tahir et al. Control Engineering Practice 67 (2017) 64–75
desired D50 crystal size by adjusting the concentration (supersaturation) desirable profiles were then tested on both the continuous reactors and
setpoint. This setpoint is then written down to the concentration MPC tuned further to optimise yield, mean particle size and span. Note that
controller which has an interval of 60 s (for DN15) and manipulates span is a measure of the width of the particle size distribution and
the feed and seed pumps. As shown in Fig. 5, the controller receives is calculated by subtracting the lower decile from upper decile and
real-time concentration predictions from a PLS calibration model based dividing the result by the median particle size.
on the FTIR spectral data as well as temperature. Fig. 7 shows the optimised temperature cooling profile, as applied
It is worth mentioning that the control structure in Fig. 5 is gener- to each of the thermocouples on the DN15 as well as the APC tracking
alised for continuous crystallisers. This structure and process model can results. The red bars represent the desired cooling profile (which is the
be programmatically reconfigured to support transfer between reactors. optimal temperature value with a margin of ±2 ◦ C as the target for
Furthermore, both cascade controllers are transferable between reactors the controller). The blue vertical bars represent the actual temperature
values achieved by the APC during a particular run along with the
by model switching/adaption techniques and deliver crystals of the
measurement uncertainty. Finally, the blue horizontal bars are simply to
desired size.
demarcate/clarify different thermocouple temperature values on Fig. 7
(they do not represent any error margins/uncertainty).
5. Temperature profile development and control Note that a similar cooling profile was also applied to the Rat-
tlesnake.
To produce crystals of the required final product quality, the cooling
profile should be optimised to ensure growth occurs within the Meta- 5.1. MPC control for reactor temperature profile
Stable Zone throughout the evolution of the reaction. In order to develop
an optimal cooling profile for Lactose in solution, a set of offline trials In the context of reactor temperature profile control, multivariate
were run in a batch stirred tank reactor (results shown in Fig. 6). The control schemes such as MPC, offer several benefits over manual/PI
70
F. Tahir et al. Control Engineering Practice 67 (2017) 64–75
temperature model, in terms of the step response from MVs to the CVs, is
shown in Fig. 10. In particular, Fig. 10(a) gives an overview of the model Fig. 12. Training data for calibration model.
structure. Then, Fig. 10(b) shows the close-up of the step responses
between the circulator temperatures (MVs) and thermocouple temper-
atures (CVs). This shows the aforementioned multivariable interactions DN15 has an impact on the temperature in the adjoining straights.
at play within the process (for example, adjusting circulator 2 affects Similarly, for Rattlesnake, the cold water flow in each of the four tubes is
the temperature at thermocouples 3–6). manipulated from a single header and is therefore coupled to the others
(as shown in Fig. 2). These considerations make the implementation
Remark 3. In comparison to other control algorithms such as PI of PI control scheme quite challenging whereas the MPC algorithm
control, an MPC based advanced control scheme is more suited to the is able to effectively deal with constraints and interactions within its
considered continuous crystallisation process. This is due to the fact formulation (see Fig. 11).
both DN15 and Rattlesnake introduce plug flow with large residence
times (1 h+) . Furthermore, as discussed above, the process involves The temperature control interval for both the Rattlesnake and DN15
actuator constraints and multivariable interactions. For instance, as was chosen to be 60 s. Fig. 11 shows the MPC based reactor temperature
can been seen in Fig. 10, adjusting a single circulator setpoint in control for the Rattlesnake unit. Notice that the MPC makes gentle
71
F. Tahir et al. Control Engineering Practice 67 (2017) 64–75
(this is shown in Fig. 12(a)). The FBRM was applied in conjunction with
temperature cycling to determine the onset of nucleation and dissolution
at each concentration.
Secondly, the wavelength range specific to Lactose (900–1200 cm−1 )
was identified and selected for calibration modelling. Fig. 12 shows the
theoretical concentrations of Lactose (36%–49%) used in the experiment
(a) and recorded spectra (b).
Fig. 14. MPC control structure for concentration. Thirdly, spectra were pre-processed using Savitzky–Golay filter. The
filter with polynomial of 2nd degree and first derivative used ±5 samples
window. Finally, using the above shown data, a calibration model was
moves across several actuators to track the product temperature setpoint trained to predict the Lactose concentration using Partial Least Squares
changes. (PLS) on the pre-processed spectra and reactor temperature.
PLS technique has been widely used in chemometrics to build
6. Control of final product quality (CQA) calibration models. The method offers reduction of variables whilst
capturing all the significant information within the entire signal range.
In a batch system, the crystallisation reaction is typically initiated PLS parameter estimation was used to obtain a calibration model.
by introducing a seed slurry or by rapid cooling to induce nucleation. The model predicts the concentration of Lactose in a solution using
Seeding is a one-time event that is significant in determining the final the FTIR spectrum as well as solution temperature. Information was
particle size distribution. High seed loading results in the rapid onset extracted from all wavelengths of a spectrum and solution temperature,
of nucleation and generation of fines. On the other hand, a low seed and stored in three latent variables. These variables explained 99%
loading in the solution may not be enough to induce crystal growth. With of variability within the training data and, at the same time, were
continuous crystallisation, the opportunity exists to adjust seed-loading correlated to Lactose concentration. Fig. 13 shows that there is a good
to control final product quality. Similarly, with very short residence match between the designed calibration model predictions and the
times, the solution may not have enough time to crystallise, whilst theoretical concentrations used in the experiment.
long residence time may affect the plug flow. Therefore, to achieve Once the calibration model was commissioned in the APC platform,
the end product with multiple desired properties, it is advantageous to the concentration control model (structure shown in Fig. 14) was identi-
manipulate different available parameters (e.g. seed and feed pumps). fied by applying process response tests on Feed and seed pumps speeds,
The APC solution uses online PAT to obtain and infer in-line respectively, as shown in Fig. 15. It can be seen that as the seed volume
measurements of crystal properties. Analytical instruments used in is increased, the concentration of Lactose in the solution decreases.
this application include FTIR spectroscopy and FBRM. Such inline On the other hand, there is a positive correlation between the feed
methods allow Real-Time Release and/or reduce the requirement for pump speed and concentration. These interactions were quantitatively
off-line product quality measurement. Moreover, inline measurements captured (using RLS) in the identified concentration control model
play a fundamental role in product quality improvement through the shown in Fig. 16. In particular, it shows the concentration response to
application of closed-loop control. step changes in the seed pump and feed pump speeds, respectively.
Here, it is worth mentioning that the developed linear models were
6.1. Real-time concentration prediction and control adapted online in terms of the dead time (pure delay). This is because
when the MPC adjusts the seed and feed rates (in order to control to the
In the APC solution, Super-saturation (concentration) is controlled concentration setpoint), the process residence time changes. Therefore,
by adjusting seed and feed loading. To this end, both DN15 and the APC scheme takes account of this by dynamically adapting the
Rattlesnake were fitted with an FTIR spectrometer at an intermediate time-delay within the concentration models for both the DN15 and
position along the reactor. Chemometric techniques were then used Rattlesnake.
to develop a calibration model to predict the Lactose concentration in
solution. 6.2. Crystal size control
A programmable experimental method was developed to allow the
generation of Meta-Stable Zone boundary information and spectral data Offline product analysis gives an advantage of being able to obtain
for calibration modelling. To minimise API consumption and devel- more reliable Particle Size Distribution (PSD) estimates. However, lack
opment time, the experimental method was implemented in a batch of instant results and inability to apply remedial changes to the process
version of the OBR. Solutions of concentration from 36% to 49% (grams is its biggest disadvantage. To this end, the proposed scheme uses
of Lactose/grams of water) were prepared for the experimental method FBRM to estimate, online, the size and count of particles in a solution
72
F. Tahir et al. Control Engineering Practice 67 (2017) 64–75
Fig. 15. Feed and seed process response tests on the Rattlesnake.
7. Conclusions
Fig. 17. MPC control structure for particle size. The paper described a transferable advanced control solution for
continuous crystallisation reactors. The solution was developed and
tested with both Cambridge Reactor Design’s Rattlesnake COBR as well
as Nitech’s DN15 COBR. The APC control was tested using Lactose as an
using scattered light. The measurements collected from FBRM can be
API, though it can easily be adapted to use other materials. Similarly,
transformed into a Chord Length Distribution (CLD), which provides
the solution is transferable across continuous crystallisation reactors.
information on size of crystals. Research shows that use of 10th,
50th and 90th percentile (D10, D50, D90) of that distribution can be The proposed solution consisted of three MPC controllers. The first
considered as relevant metrics for defining size of crystals. Furthermore, one was a generic block, which is applied to control reactor temperature
sufficient representation of crystal size can be obtained from median in order to maintain the desired cooling profile. The other two con-
chord length (D50) alone. Therefore, this measurement was chosen to trollers were used, in a cascade manner, to drive the API concentration
be controlled by the third APC controller, which used concentration and particle size (D50) towards the desired setpoints. Both of the
of Lactose within solution as an actuator (as shown in the controller controllers incorporate PAT data, which allow real-time monitoring
structure in Fig. 17). and control of the product’s quality as well as rejection of any process
The FBRM was installed close to the outlet of each crystalliser. This disturbances such as reactor fouling and API impurities. In particular,
position meant that the solution was analysed near the end of the spectral data collected from FTIR was used to build a calibration model,
cooling profile, where crystallisation should be at its final stage. Fig. 18 which predicted the concentration of API in the solution. Concentration
shows the control results with the cascade strategy, where a change in of the API was then controlled by APC through manipulation of seed and
setpoint of the median chord length (D50) prompts steady reduction of feed flows. Moreover, a successful control of the median Chord Length
73
F. Tahir et al. Control Engineering Practice 67 (2017) 64–75
Distribution (D50), obtained from FBRM, was achieved by manipulating ∙ CLD: Chord Length Distribution
concentration of the API in real-time. ∙ COBR: Continuous Oscillatory Baffled Reactor
As mentioned above, the control scheme can easily be extended ∙ CQA: Critical Quality Attribute
to accommodate other APIs. It would then be possible to interchange ∙ CSTR: Continuous Stirred-tank Reactor
the API specific calibration and particle size models, according to ∙ CV: Controlled Variable
the manufacturing needs and production schedule. Furthermore, the ∙ FBRM: Focused Beam Reflectance Measurement
structural differences between various reactors can be compensated by
∙ FTIR: Fourier Transform Infrared
modifying/adapting time delays within the models in real time.
∙ MPC: Model Predictive Control
∙ MSZ: Meta-Stable Zone
Acknowledgement
∙ MV: Manipulated Variable
∙ OBR: Oscillatory Baffled Reactor
The authors gratefully acknowledge funding support from Innovate
UK under the project ‘MOPPS: Made To Order Processing Plants’ (grant ∙ OE: Output Error
number: 101334). ∙ PAT: Process Analytical Technology
∙ PLS: Partial Least Squares
Appendix. List of acronyms ∙ PRBS: Pseudo Random Binary Sequence
∙ PSD: Particle Size Distribution
∙ APC: Advanced Process Control ∙ QbD: Quality-by-Design
∙ API: Active Pharmaceutical Ingredient ∙ RLS: Recursive Least Square.
74
F. Tahir et al. Control Engineering Practice 67 (2017) 64–75
References Mayne, D. Q., Rawlings, J. B., Rao, C. V., & Scokaert, P. O. M. (2000). Constrained model
predictive control: Stability and optimality. Automatica, 36(6), 789–814.
Cambridge reactor design. http://www.cambridgereactordesign.com/continuous-crystall McGlone, T., Briggs, N. E. B., Clark, C. A., Brown, C. J., Sefcik, J., & Florence, A. J. (2015).
isation.html#. Oscillatory flow reactors (OFRs) for continuous manufacturing and crystallization.
Chiu, T., & Christofides, P. D. (1999). Nonlinear control of particulate processes. AIChE Organic Process Research and Development , 19(9), 1186–1202.
Journal, 45(6), 1279–1297. Nitech solutions. http://www.nitechsolutions.co.uk/products/dn15-standard/.
Forgione, M., Birpoutsoukis, G., Bombois, X., Mesbah, A., Daudey, P. J., & Van den Hof, P. Orkoulas, G., Kwon, J. S., Nayhouse, M., & Christofides, P. D. (2014). Crystal shape
M. J. (2015). Batch-to-batch model improvement for cooling crystallization. Control and size control using a plug flow crystallization configuration. Chemical Engineering
Engineering Practice, 41, 72–82. Science, 119, 30–39.
Georgieva, P., Meireles, M. J., & Feyo de Azevedo, S. (2003). Knowledge-based hybrid Perceptive engineering limited. http://www.perceptiveapc.com.
modelling of a batch crystallisation when accounting for nucleation, growth and Powell, M. J. D. (1983). ZQPCVX a FORTRAN subroutine for convex quadratic programming .
agglomeration phenomena. Chemical Engineering Science, 58(16), 3699–3713. Department of Applied Mathematics and Theoretical Physics, University.
Goldfarb, D., & Idnani, A. (1983). A numerically stable dual method for solving strictly Qin, S. J., & Badgwell, T. A. (2003). A survey of industrial model predictive control
convex quadratic programs. Mathematical Programming , 27 (1), 1–33. technology. Control Engineering Practice, 11(7), 733–764.
Goodwin, G. C., & Sin, K. S. (2009). Adaptive filtering prediction and control. New York, NY, Siddique, H., Brown, C. J., Houson, I., & Florence, A. J. (2015). Establishment of a
USA: Dover Publications, Inc.. continuous sonocrystallization process for lactose in an oscillatory baffled crystallizer.
Kwon, J. S., Nayhouse, M., Christofides, P. D., & Orkoulas, G. (2014). Modeling and control Organic Process Research and Development , 19(12), 1871–1881.
of crystal shape in continuous protein crystallization. Chemical Engineering Science, Siddique, H., Raval, V., Mabbott, F., Mack, J., Tahir, F., Krzemieniewska, K., Houson, I.,
107 , 47–57. Mercer, E., Florence, A. (2017). A direct design approach for ‘‘Dial a Particle size
Li, M., Mhaskar, P., Shi, D., El-Farra, N. H., & Christofides, P. D. (2006). Predictive control capability’’ in continuous crystallisation of Lactose, submitted for publication.
of particle size distribution in particulate processes. Chemical Engineering Science, Tahir, F., & Jaimoukha, I. M. (2013). Robust feedback model predictive control of
61(1), 268–281. constrained uncertain systems. Journal of Process Control, 23, 189–200.
Maciejowski, J. M. (2002). Predictive control with constraints. Essex, England: Prentice Hall. Zhao, L., Raval, V., Briggs, N. E. B., Bhardwaj, R. M., McGlone, T., Oswald, I. D. H., et
Mack, J. (2013). Pat based advanced process control and monitoring. In ISPE annual al. (2014). From discovery to scale-up: 𝛼-lipoic acid : nicotinamide co-crystals in
meeting . a continuous oscillatory baffled crystalliser. Crystal Engineering Communications, 16,
5769–5780.
75