(670c) Hypothesis Testing and Process Understanding in the Agrochemical Industry: A Case Study Utilizing First-Principle Modeling with Limited Measurable Data

First-principle models are often perceived as complex and challenging to calibrate. These models require specialized expertise for proper formulation and application, and the specific systems of interest may not be sufficiently understood for accurate industrial modeling. Crystallization serves as a relevant example; while there is a solid foundational knowledge of its fundamental physics, the capability for detailed modeling and simulation of complex, realistic systems remains limited, although it is improving. Even when modeling is feasible, economic constraints often hinder the precise determination of all necessary parameters. Nonetheless, it is possible to develop "good enough" models that capture the essential features, trends, and behaviors of the system in question. Utilizing such models to interpret and design experiments can significantly enhance our understanding, which is crucial for developing more efficient processes. The chemical engineering community, particularly within the crystallization sector, has successfully applied this approach in the past, as demonstrated by advancements in solid-state deracemization.

In this presentation, we illustrate how industrial applications can derive substantial benefits from this modeling approach. We will present a case study from Bayer's current agrochemical portfolio, focusing on optimizing a multicomponent, multiphase system designed to produce a key active ingredient. This project also facilitates technology transfer across various production sites.

The system consists of a primary reaction aimed at synthesizing the desired active ingredient, followed by a crystallization phase. This crystallization process is intricately influenced by the supersaturation levels established during the initial reaction stages and the subsequent cooling of the system. The optimization challenge stems from three key factors: first, the process is structured in a "telescopic" fashion, effectively minimizing separation between synthesis steps; however, this design can lead to the carryover of impurities that may react and accumulate in the final stage. Second, the solvent composition—a ternary mixture—exhibits variability from batch to batch due to both technological and commercial considerations. Lastly, accurately scaling the hydrodynamic conditions of the system is crucial, as it can significantly affect both the reaction and crystallization processes through mechanisms such as secondary nucleation and agglomeration.

Given the constraints of conducting an extensive experimental lab campaign, we focused on modeling the critical aspects of the chemistry and crystallization processes. This model, based on our current understanding of the system, will be used to investigate how variations in operating parameters—such as temperature, feeding rate, and solvent mixture composition—affect particle size and shape distribution, yield, and productivity. Additionally, we have implemented a simplified non-ideal mixing model that allows us to computationally reproduce certain design choices, such as the location of feeding, and assess their impact on overall crystallization. Following the simulations, we will conduct targeted lab-scale experiments to validate the trends observed, thereby confirming or challenging our initial hypotheses regarding the system.