**Coupling with Fluid Mechanics**

As previously mentioned, due to the progress of computer efficiency, memory and rapidity and the availability of both commercial and open-source codes of fluid mechanics, these advanced numerical techniques are more and more frequently applied to traditional processes, which are, thus, revisited and improved. Zeng et al. [8] study a particular unit, the turbo air classifier, which is used to separate powders into coarse and fine. Based on the governing equations and using a commercially available software, they study the influence of operation parameters, such as rotation speed and

air volume. Studies are performed for multiple particles. Simulation and experimental results are favorably compared. Cheng et al. [9] also use a computational fluid dynamics code to better understand bubble columns, which are frequently used in process industries and whose behaviour is complex and not totally predictable. They use the Euler–Euler approach, consisting in obtaining mean local properties to describe the bubble plume, and compare their simulation results to laboratory experiments. They reveal the correlation of operating conditions with the gas mixing and plume oscillation period. The commercial code of CFD is the same as used by [8].

## **Studies of Given Processes**

It is natural and usual that some given processes are studied by researchers who either try to promote specialized models or improve their behavior by examining operation parameters. Many studies can be gathered in this category. Yang et al. [10] study an in-situ combustion method that can be used for bitumen sands or heavy oil production. Their numerical study deals with a given block of a Chinese oil field. They analyze the influence of production (hu ff and pu ff rounds, air injection speed and air injection temperature) and geological parameters (bottom water thickness, stratigraphic layering, permeability ratio and formation thickness). They are thus able to extract the main operation parameters. Shakeel et al. [11] compare the industrial production of formaldehyde using two di fferent catalysts. They use a commercial steady-state process simulator to simulate the process and deduce the advantages and drawbacks of these configurations. Chinh et al. [12] study a nitrogen gas separator. This is a fixed bed operated by pressure swing adsorption. Thus, it is an appropriate subject for modeling and simulation with its dynamic and sequential operation, but the authors also compare their results to a laboratory pilot plant. The model, composed of partial di fferential and algebraic equations, is fully described together with its operating parameters. Chen et al. [13] study the production of butyric anhydride by means of a single reactive distillation column. Although this process has already been described in the literature, its applications are not numerous due to many operation di fficulties and lack of generality. Moreover, the authors have replaced a two-column process with a single one by using the internal circulation of acetic anhydride. They show that this can be extended to similar reactive distillation two-column processes. The simulation is performed by a commercially available steady-state process simulator by which they optimize the process. Finally, they perform an economic analysis of the novel process. Marecka-Migacz et al. [14] study a membrane process—more specifically, ceramic nanofiltration for the separation of succinic acid aqueous solutions. They provide a detailed model of nanofiltration for ion transport, taking into account convection, di ffusion and electromigration. They compare the results obtained by their elaborate model to those issued from more standard approaches. In their case, all species are considered, as well as solutes, ions and solvent, pH-regulating solutions and water. The conditions of use of their more complete model are emphasized. Experiments are also performed for comparison. Nguyen et al. [15], again, study a membrane process of ultrafiltration of protein. The model is simpler than in [14]. Their objective is to perform an economic assessment of the process and, for this purpose, their focus is related to the design parameters. With the latter being numerous, they perform the optimization by means of a genetic algorithm in a black-box manner, although they rely on their analytical model.
