Simulation models have been developed for an existing I 100 litre and a 360 m(3) draft tube baffle (DTB), crystallizer. For this purpose, various compartmental models are employed. The resulting simulations provide a general indication of the importance of (i) the use of real dissolution kinetics as opposed to the assumption of complete dissolution for the fines removal system of a DTB crystallizer; (ii) taking into account the hold-up of the annular zone; (iii) compartmentation of the crystallizer main body; and (iv) the modelling of internal classification. For the generation of optimal crystallizer designs, mathematical optimization techniques are used that allow a flexible formulation of constraints and an objective function. The design problem considered here involves a DTB crystallizer for the crystallization of ammonium sulphate from water with a specified production capacity. A transparent solution strategy is presented, which is used to quantitatively explore three specific cases, including a multistage system, all involving the same economic objective. The results, which are specific for the chosen combination of product specifications, design variables and operating conditions, indicate that the production of 94 kton year(-1) with a target median crystal size of 2 mm or less in a single DTB crystallizer only leaves the propeller frequency as an active design variable to achieve this median size. A draft tube (DT) crystallizer would therefore give a more optimal solution in this case. Only when crystal sizes beyond 2 nun are desired can other design variables such as the fines removal rate and product residence time be exploited to optimize the design. Finally it was found that production in multiple crystallizers in parallel with a combined production capacity can be economically beneficial.