Development of a Model for a Reaction Mixing Pump

Stirred tank reactors are widely used in various chemical processes due to their efficient mixing properties, which increase mass and heat transfer. However, there is now a promising alternative known as reaction mixing pumps (RMPs), which combine mixing and transportation of reaction media, resulting in process intensification. Furthermore, the design of the pumps enables very high internal circulation flows that are beneficial for mixing sensitive reaction systems. Despite the potential benefits of RMPs, there has been limited research conducted on this specific type of reactor.

In order to use RMPs as reactors in future applications, models are needed that can describe the behavior of the RMP as a reactor and thus predict yield and selectivity for any reaction network. In principle, these parameters can be obtained from residence time measurements. However, residence time measurements are expensive and are usually not performed by the pump manufacturer. This means that in the context of process development, this information cannot be used to select and size a suitable RMP. Therefore, we have developed a workflow that provides a reliable reactor model of the RMP based on readily available pump information.

We chose a compartment model to describe the RMP as reactor. However, this compartment model initially contains unknown parameters that reflect the internal flows in the pump. These internal flows are estimated with a fluid dynamic model of the pump based on the conservation of angular momentum. This fluid dynamic model requires parameters that describe the friction losses in the pump. These friction parameters are also initially unknown. In our presentation, we will show that these parameters can be well estimated from the pump's characteristic curves. The characteristic curves are well suited for this purpose as they are typically provided by the pump manufacturer or can be easily measured. In the next step, the internal flow rates predicted by the fluid dynamic model can be used as input for the compartment model. Comparisons with residence time distribution measurements show that the compartment model parameterized with our approach can predict the behavior of RMPs as a reactor.

Using the compartment model and supported by the residence time measurements, we were able to find operating parameters (i.e., pump speed, pressure increase) for which the behavior of the RMPs comes close to that of an ideal continuous stirred tank reactor. Results from reactive experiments with mixing sensitive reactions confirm these findings.

We show that our model approach is very useful to describe the behavior of the pump and to optimize the operating parameters. Furthermore, we present the RMP as an attractive alternative to stirred tank reactors for achieving high yields in fast, mixing-sensitive reactions.