Salt Crystallization on a 1 m 3 Scale: From Hierarchical Design to Pilot Plant Operation Marcelo M. Seckler,* ,† Marco Giulietti, ‡ Andre ́ Bernardo, ‡ Silas Derenzo, § Efraim Cekinski, § Andre ́ Nunis da Silva, †,§ Herman J. M. Kramer, ∥ and Max Bosch ⊥ † Department of Chemical Engineering, University of Sã o Paulo, Av. Prof. Luciano Gualberto, travessa 3, 380, Butantã , 05508-010, Sã o Paulo, SP, Brazil ‡ Department of Chemical Engineering, Federal University of Sã o Carlos, Rodovia Washington Luiz km 235, 13560-470, Monjolinho, Sã o Carlos, SP, Brazil § Institute of Technological Research of the State of Sã o Paulo, Av. Almeida Prado 532, Butantã , 05508-901, Sã o Paulo, SP, Brazil ∥ Department Process & Energy, Delft University of Technology, Leeghwaterstraat 44, 2628CA, Delft, The Netherlands ⊥ Refinaria Nacional de Sal, Salinas Ponta do Costa s/n, 28901-970, Cabo Frio, RJ, Brazil ABSTRACT: The synthesis of solution crystallization processes is a complex task that often leads to multiple process options. In order ensure design reproducibility and reliability, a hierarchical design procedure has been proposed. The procedure has compared favorably to the currently accepted procedure because the number of design decisions is more evenly distributed throughout the design levels. The procedure is based on the work of Bermingham (A design procedure and predictive models for solution crystallization processes. Ph.D. thesis, Delft University of Technology, 2003), but recourse to sophisticated phenomenological models is avoided. Instead, experimental information, heuristics and qualitative theoretical considerations are used to cope with systems for which fragmentary information is available, since such is the situation most commonly found in industrial practice. Its applicability has been demonstrated in the design of a sodium chloride crystallization process on a 1 m 3 scale. Analysis of the pilot unit operation has led to the identification of improved design criteria related to process control, temperature elevation in the recirculation loop, and crystals washing. ■ INTRODUCTION Processes involving crystallization from solutions are currently designed by a procedure preconized by several authors. 2−4 This procedure has been nicely complemented by design procedures that address the design needs of specific systems within the field of crystallization from solutions. Such is the case for systems involving solid solutions, 5 for multicomponent systems that require fractional crystallization, 6 and for polymorphic systems. 7 In addition, procedures have been developed for the integrated design of crystallization and downstream processing operations such as filtration, washing, dewatering, recrystallization, and drying. 8 The integration of design and control has also been proposed upon considering that crystallization involves highly nonlinear steps and is applied in processes comprising several unit operations and recycle streams. 9 Recently, a generic approach to design has been proposed, which decomposes the process into tasks that represent fundamental physical events. 10,11 This procedure is promising for process intensification, but it needs further development before it can be applied. In spite of these developments, the synthesis of crystallization processes remains a complex task that requires many hypotheses and often leads to multiple solutions, so that the resulting design largely depends on the designer’s experience. Therefore, systematic design approaches are still needed in order to simplify and improve the quality of design. It is desirable to develop design procedures that comply with the following requirements: (i) a short development time, (ii) little experimental effort, and (iii) reproducibility, i.e., the procedure should include traceable arguments and design decisions. Douglas 12 has proposed a hierarchical design procedure that consists of a number of design stages. In each stage a limited number of aspects are considered, thereby simplifying the processes of formulating hypotheses and making decisions. The procedure, which has been developed for vapor−liquid systems, has been adapted to vapor−liquid−solid systems by Rajagopal and co-workers. 13 A similar procedure has been derived later for crystallization processes, 1,14−16 with a focus on a better prediction of the final product quality through detailed mathematical modeling of the crystallization phenom- ena. The disadvantage of this approach is the difficulty in obtaining the required experimental information, as laboratory data are difficult to translate to the industrial scale. Therefore, in this work a hierarchical design method has been proposed that precludes the recourse to phenomenological models. Instead, it is largely based on fragmentary experimental information, heuristics, and theoretical considerations. This simplified procedure is applicable to a large number of crystallization systems, for which the crystallization behavior is only partially known. The advantages of the proposed procedure have been highlighted by a comparison with the Received: September 28, 2012 Revised: December 18, 2012 Accepted: February 24, 2013 Published: February 25, 2013 Article pubs.acs.org/IECR © 2013 American Chemical Society 4161 dx.doi.org/10.1021/ie302657n | Ind. Eng. Chem. Res. 2013, 52, 4161−4167