chemical engineering research and design 89 (2011) 995–1005 Contents lists available at ScienceDirect Chemical Engineering Research and Design journal homepage: www.elsevier.com/locate/cherd Rapid online calibration for ATR-FTIR spectroscopy during batch crystallization of ammonium sulphate in a semi-industrial scale crystallizer Somnath S. Kadam a,* , Ali Mesbah a,b , Eric van der Windt c , Herman J.M. Kramer a a Process and Energy Laboratory, Delft University of Technology, Leeghwaterstraat 44, 2628 CA, Delft, The Netherlands b Delft Center for Systems and Control, Delft University of Technology, Mekelweg 2, 2628 CD, Delft, The Netherlands c Bruker Optics B.V., Oostsingel 209, 2612 HL, Delft, The Netherlands abstract The knowledge of solute concentration throughout a batch crystallization process is essential from process control perspective. Despite the progress in process analytical technology (PAT), there still exist several challenges for online measurement of solute concentration at industrial scale. In this study, concentration monitoring was realized using attenuated total reflectance Fourier transform infrared (ATR-FTIR) spectroscopy at lab as well as semi-industrial scale. Applications of the calibration model developed at lab scale for the measurements at the semi-industrial scale however resulted into strongly biased concentration predictions, caused by the differences in the curvature of fiber optics and the uneven thermal expansion of the probe. Therefore an alternative rapid online calibration method was developed during the start-up phase of the process. With this method, the time required for developing a working calibration model for concentration monitoring during crystallization of ammonium sulphate in a semi-industrial scale draft tube crystallizer has been reduced approximately by 90%. With the help of simultaneous concentration and crystal size distribution measurements at semi-industrial scale, the descriptive capability of the model was improved due to better kinetic parameters. © 2010 The Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved. Keywords: Batch crystallization; Semi-industrial scale; Concentration monitoring; ATR-FTIR spectroscopy; Calibra- tion; Parameter estimation 1. Introduction Crystallization is an important unit operation which deter- mines the product quality in the pharmaceutical, specialty, fine and agrochemicals industry. Generally, the crystal quality is defined in terms of the purity, size and shape distribution, polymorphic fraction, etc. (Mersmann, 2001). These properties are highly dependent on the operating conditions in the crys- tallizer. A small variation in these properties may affect the efficiency of the downstream processing units and the prod- uct performance (Braatz, 2002; Wibowo et al., 2001). Hence, an effective control over these properties and in turn the crystal- lization process is essential. In recent years, with the developments in the computing power and the in situ monitoring of the process variables like ∗ Corresponding author. Tel.: +31 015 2786628; fax: +31 015 2782460. E-mail address: s.s.kadam@tudelft.nl (S.S. Kadam). Received 8 September 2010; Received in revised form 15 November 2010; Accepted 22 November 2010 crystal size distribution (CSD) and concentration, two distinct control approaches have evolved. One of these approaches (model-free) requires determination of a supersaturation pro- file (generally experimental) which seeks a trade-off between crystal quality and process productivity. The supersatura- tion profile determined in this way is always near-optimal. Once the supersaturation profile has been determined, a con- trol strategy is devised which maintains the desired relation between the system states, viz. concentration and tempera- ture, along the predefined supersaturation profile (Fujiwara et al., 2005; Nagy et al., 2008; Zhou et al., 2006). The other approach (model-based) relies on the model developed with the population balance (Hulburt and Katz, 1964), conservation laws and kinetic expressions (Braatz, 2002; Mesbah et al., 2010; Nagy, 2009). This model is used in an optimization framework 0263-8762/$ – see front matter © 2010 The Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved. doi:10.1016/j.cherd.2010.11.013