(394e) Improvement of Design Assessment Methodologies for Melt Static Crystallization Process

The use of melt crystallization as a separation and purification technology   has proved to be an efficient process for the treatment of organics. Its development accelerated in the fifties up to its current industrialization status. Nowadays, about 50 major compounds are treated by melt crystallization. As for any process design issue, the choice of melt crystallization is the result of a selection process driven by key performance indicators such as relative efficiency, comparative costs, improved safety or specific performance features not allowed by other types of unit operations. As examples of melt crystallization units one can discern various cases such as a crystallization unit used as a polishing unit downstream a distillation column in order to reduce the overall capital and operational expenditures, separation of compounds with closed boiling points and therefore not easily achieved by distillation, possibility to handle under safe conditions some sensitive compounds that may be degraded if heated up…

The solid-liquid phase diagram (SLE) is a good way to very roughly assess about the applicability of a separation by melt crystallization. It is however not sufficient to assess about its design or potential performance. Indeed any crystallization from the melt is not only resulting from thermodynamic controlled features. The crystallization from the melt is intrinsically a complex phenomenon involving multi-physic interactions:

-          Thermodynamic s as ruled per SLE

-          Crystallization kinetics

-          Multiphase interaction between the formed solid  and the liquid phase

-          Transient heat exchange performance and thermal gradient impacting the two former points

Among the realization of a melt crystallization based process, various declinations or modes are possible. One can consider the continuous and the discontinuous types of crystallizations from the melt. A continuous type can typically be simplified as for example a scrapped heat exchanger coupled with a solid/liquid separator such as a frame filter or a centrifuge. A discontinuous mode can be embodied as batch process where the product to be purified is either kept under natural convection (melt static crystallization (MSC)) or flowed under forced convection either as an ascending phase or has a trickling film (both are referred to as dynamic crystallizations). Each mode implies that the key driving factors implied may be slightly different and that their relative impacts on the physical interactions lead to different consequences over the performances achieved by each process variant. 

Within the frame of an industrial process design aiming at comparing a melt crystallization workshop with other unit operations, all these factors must be considered and integrated all together. This is also true while comparing different melt crystallization variants. This is yet the only way to get a representative view of an expected process performance. As such and as long as the crystallization from the melt is considered any process assessment is a far less direct route than what can be achieved by well established process simulators used for the design of a distillation column for example.

For the specific case of MSC Franke proposed a short cut approach based on a modification of the SLE for sizing the crystallization plant. This black box approach has been introduced for a p-xylene case only and appeared to be restrictive as it is limited to processes using a single crystallization stage. For more complex process designs such as fractional crystallization combining multiple stages, this approach cannot be generalized. The reasons lie within the mechanics governing each stage performance. A MSC stage is divided into a series of consecutive steps:

-          Crystallization

-          Draining

-          Sweating

-          Melting

On the first hand the performance of each stage is linked, first to the feed composition, then to the management of the crystallization step which affects the amount of impurities hold up in the solid. On the second hand the final performance depends of the sweating step which is used to correct the effect of this hold up; the sweating management can improve the stage performance to various extents. The achieved efficiency results not only from the definition of the heating applied to the sweating step but also from the state inherited at the end of the crystallization step. Large interferences may therefore happen, the latter being related to the quality of the feed crystallized, the thermal management of the crystallization, extend of crystallization achieved and the thermal management of the sweating. The observed effects are resulting from the underlying consequences of thermodynamics, kinetics and rheology.

On the basis of the approach introduced by Franke, this contribution proposes an improvement of the methodology for MSC conceptual/preliminary design assessment. This approach is declined for eutectic forming systems and solid solution forming systems as per the classification proposed by Matsuoka and Fukushima. The introduced shortcut method is based on a modeling of the impurity content in the solid based on a multi-parameter correlation including feed quality, crystallization and sweating extend. The robustness of the approach is evaluated by comparison of the obtained results with industrial cases taken from Fives portfolio for MSC fractional crystallization processes.