CD Laboratory for Multi-Scale Modeling of Multiphase Processes

In so-called multiphase processes, several separate phases - gases, liquids or solids - interact with each other, e.g. through reactions. The exact processes in such reactors are now to be made calculable and thus designable.

Multiphase processes, such as those used in fluidised bed reactors (biomass reactors, reactors for polymer production, etc.), are considered to be extremely effective process methods, as they ensure efficient mixing of the phases and a high contact rate. Until now, however, the design of such reactors has relied on empirical findings based on previous experience, analytical considerations and trial and error, as the insufficient computer capacities did not allow the underlying processes to be modelled. An important reason for the complexity of such processes is the drastically different scales involved, which are, however, directly dependent on each other. For example, multiphase reactors are typically several metres high, but the mixing and reaction processes that take place in them are significantly influenced in detail by the behaviour of microscopically small particles and particle conglomerates - for example, by the collisions of individual particles with each other or with the reactor wall, or by flow resistance between the phases. Another challenge is the massive discrepancies in the temporal dimensions involved. The sequence of such reaction processes spans from the processes between the individual particles, which take place in fractions of a second, to the duration of the entire reaction process of several hours.

The aim of the CD Laboratory's research is therefore to make these enormous spatial and temporal differences in scale calculable and thus plannable using new methods for simulating the processes on several scale levels (multi-scale modelling). In this way, multiphase reactors are to be removed from the trial-and-error level in future and become specifically designable.

The research approach consists of three levels, which can simulate the different dimensional levels separately and then merge them. At the spatial level, small-scale local phenomena can be efficiently and simultaneously resolved in a coarse-scale simulation using a "magnifying glass" concept. Subgrid models take into account the influence of unresolved small-scale effects; they also depend on the macroscopic properties of the multiphase process. At the temporal level, a process model takes into account the wide range of time scales involved by reducing the three-dimensional multiphase flow to a conditional random process. The aim of the application of such random processes is to specify probabilities for the occurrence of future events even if only a limited history is known. The latter approach is expected to increase the efficiency of numerical simulation by several orders of magnitude. Finally, these novel concepts for multiphase modelling are verified and validated by analytical and experimental investigations.

This should not only make it possible to design new reactors, but also make it easier to analyse faults in the event of process failures.