Research

• Fluid Dynamics, Atmospheric Dispersion Modeling
• High-Order Coupled Cluster Method
• Solubility Prediction, Crystal Structure Optimization

Dynamics of reactive fluids, atmospheric dispersion modeling

For problems dealing with reactive fluids or within the field of dispersion modeling one is confronted with numerous working materials, starting with rules of thumb up to complex numerical simulations.

Many practical applications rely on standards based on empirical investigations, while precise numerical simulations are limited to specific models.

Unfortunately not the minority of investigations suffer from a lack of tolerance/error-influence investigation resulting in a questionable reliability of the conclusions. Meanwhile well elaborated Navier-Stokes-based numerical models and corresponding software is available for problems related to reactive fluids or to dispersion modeling. Nevertheless, even in scientific publications inaccurate initial conditions are selected, the motivation for the choice of the underlying model is vague or oversimplified.

Especially a sufficient discussion of the bandwith of hypothetic events, an analysis of tolerances or a discussion of significance and relevance is often avoided. Here the impact of errors and the probability distribution of the involved variables will be included in our numerical investigations via Monte-Carlo-Algorithms. This is of pronounced relevance, if temporarily confined and local properties are used for risk calculations (PRA), while only probability distributions of influence factors are known and the validity of mean values is not sufficient.

• Applicated models: Gaussian dispersion modeling, empirical/dimensional-analytic models for dense gas dispersion (screening, close-range, according to VDI 3783 I+II), Lagrangian-particle trajectory models (long-range), CFD-RANS or CFD-LES simulations (OpenFoam, complex boundary conditions),
• Consideration: correlations of governing variables, dedicated (plausible) probability distribution functions,
• Coupling of numerical solvers with Monte-Carlo-Algorithms.

High Order Coupled Cluster Methode

In cooperation with the Institute of Theoretical Physics (University of Magdeburg, condensed matter theory II)

The Coupled Cluster Method (CCM) in one of a few universal ab inito quantum many body methods originating from quantum chemistry (used for electron correlation), meanwhile with many applications in quantum physics.

Here a description of the CCM appropriate for the determination of expectation values of quantum spin systems is used. Quantum spin systems are abstract lattice representaions of solid body structures, in which the lattice sites carry a localized magnetic moment (spin); thus the magnetic properties are investigated (based on the Heisenberg-model).

For our calculations we use a well elaborated C++-code for the calculation of:

• properties like ground state energy and order parameter, derived quantities like spin-stiffness and susceptibilities and additional the singulett-triplett spin excitation (spin-gap),
• application of collinear and non-collinear model states, investigation of unfrustrated ans frustrated spin systems in arbitrary dimension,
• calculation without or with an external magnetic field,
• application of a well elaborated and tested C++-code for the calculation up to very high orders of approximation on cluster-computer or on high performance computing systems (HPCS).

Solubility Prediction, Structures of Molecular Crystals

(Formerly) In cooperation MPI for Dynamics of Complex Technical Systems Magdeburg (molecular simulations and design (MSD), physical and chemical foundations of process engineering (PCF))

Many chemical engineering problems rely on processes in liquid phases (liquid substances or mixtures). Among others: separation processes, catalytic reactions, crytallizations and transport problems.

It is useful to be able to understand and to predict the thermodynamic behaviour within a theoretical description, especially for optimization purposes and solvent (mixture) selection.

The methods used are molecular-dynamic simulations, group contribution methods as UNIFAC and COnduktor like Screening MOdels for Realistic Solvation (COSMO-RS, COSMO-SAC und dreived versions). Molecular-dynamic simulations are able to describe the interactions (here limited to a quasiclassical picture) of an ensemble of molecules. By investigating the trajectories and different (temporary) bond formations, conclusions on characteristic molecular interaction behaviour and the corresponding probability can be gained. Summarizing, the combination of methods are used for theoretical investigation of experimentaly observed characteristics, for derivation of optimizations to experimental procedures or for possible explaination of observed behaviour.

Here, the methods are used for:

• Solubility predictions of binary and multinary mixtures, solvent screening, and selective crystallization process optimization,
• investigation of methodological limitations, applicability to selected cemical mixtures,
• Optimization of selective crystallization processes and related engineering methods in liquid phases,
• simulation of molecular interactions for a microscopcal description and validation of experimental/theoretical hypotheses.