Here, you can find the different research subjects from the different phases of my scientific career.
My research group Materials Informatics is concerned with data-driven approaches for the description and integration of novel tailor-made materials. Modern materials research requires an integrative and multidisciplinary approach, which increasingly relies on methods from mathematics and computer science in addition to traditional approaches from chemistry, physics and engineering. In particular, machine learning and the evaluation of "big data" are essential for tomorrow's materials research and related engineering sciences. The development of strategies for materials discovery and development are therefore the focus of Materials Informatics.
I thank for the financial funding of the Dresden Center for Intelligent Materials (DCIM) by the Free State of Saxony and TU Dresden. I kindly thank all partners for their support.
Based on my research on active-passive membranes for cell filtration during my PhD phase, several projects in three branches evolved. Especially the numerical investigations based on the Normalized Extended Temperature Expansion Model lead to deeper insights into the role of stiffness pairing in composites.
- Biological structures inspire 1D and 2D setups with distinct active-passive material pairing. The methods from continuum mechanics can be applied to gain deeper insights into these systems.
- The multisensitivity and logic branch is based on continuum mechanical understanding of the combined mechanical reaction of active materials to non-mechanical stimuli.
- Smart structures that use the unique advantages of tailored stiffness pairing can be derived.
The research in this phase was conducted at first from the DFG-GRK 1865 (project extension due to parental leave), later by the TU Dresden as a part of the Chair of Mechanics of Multifunctional Structures (Prof. Wallmersperger). I like to thank all cooperation partners for their support.
To allow the simulation of active layer systems composed of active hydrogels and passive materials, a closer look at the material properties has to be taken. The current project focusses on the normalized swelling behavior of hydrogels under arbitrary stimulus. With this description and correct data for the swelling dependent Young's modulus, arbitrary layer systems can be modelled and simulated.
In course of this project, I finished my PhD thesis entitled "Modellierung und Simulation des Verhaltens von durchströmten schaltbaren Membranen". It is available for download here.
The research was funded by the DFG in the framework of the Research Training Group "Hydrogel-based Microsystems" (DFG-GRK 1865). My thesis supervisor was Prof. Thomas Wallmersperger (Chair of Mechanics of Multifunctional Structures). I like to thank all cooperation partners for their support.
Selectivity in membrane permeation is important for cell functioning in nature. In technology, the counterpart filtration is achieved on the level of particle rejection or separation of salts in processes like ultrafiltration or reverse osmosis. Natural selectivity goes hand in hand with gating, the ability of organisms to manipulate membrane selectivity and transport by multiple means.
The transporting structures are called channels (with selectivity) or pores (without selectivity). Stimuli regulating the "openness" of gates in a channel can be chemical, electrical, thermic or mechanic. The goal of the project is to analyse and duplicate this ability for various use in technology like chemical computing, filtration or sensing.
This project is funded by the Deutsche Forschungsgemeinschaft DFG in the framework of the Excellence Initiative programme Support the best (Institutional Strategy "The Synergetic University") at TU Dresden. The numerical part of the integral project is supervised by Prof. Thomas Wallmersperger (Chair of Mechanics of Multifunctional Structures), while experiments are carried out by the group of Prof. Andreas Richter (Chair of Polymeric Microsystems) in the cfaed Cluster of Excellence. A further cooperation partner is the Research Training Group Hydrogel-based Microsystems (DFG-GRK 1865). I like to thank all cooperation partners for their support.
Biological membranes made of phospholipids are crucial in biological systems. They separate organisms into cells and the cell plasma into compartments. Polymeric membranes play an important role in process engineering when it comes to filtration. This work focusses on simulation of membrane transport processes through both biological and polymeric membranes. Four different classes of modelling were identified.
- Membrane-current-models such as the Hodgkin-Huxley formulation are used to calculate the electric current and voltage measured in experiments. They are fitted to experimental results with phenomenological parameters.
- Parametric models are derived from thermodynamics and geared to obtain permeation parameters while not explicitly regarding the membrane.
- Continuum mechanical models treat the compartments and membrane as separate numerical regions and solve the chemo-electrical multi-field problem.
- Micro- and nanofluidic models work on microscopic basis and treat the molecule transport according to the dominating electromagnetical interactions.
The Hodgkin-Huxley model and the continuum field formulation were implemented in Matlab. The first was discretised with Runge-Kutta fourth order method, the continuum description of migration-diffusion and Poisson equation was discretised by finite element method (FEM) in space and time-discontinuous Galerkin (TDG) in time. Simulation experiments were performed for
- simple diffusion
- diffusion through semi-permeable membranes (Donnan potential)
- transduction of action potentials in nerve axions, comparable to Hodgkin and Huxley
Experiments show the result, that even on the scale of membrane transport, the continuum field assumption holds for low concentrations. For physiological concentrations, further research and refinement considering the numerical solution is required.
The research was conducted at TU Dresden and supervised by Prof. Thomas Wallmersperger (Chair of Mechanics of Multifunctional Structures) and Prof. Andreas Richter (Chair of Polymeric Microsystems). I am very grateful for their support.
Ionic Polymer Metal Composites (IPMC) belong to the class of electro-active polymers (EAP). Due to their structure, they perform bending when an electric voltage is applied between the electrodes. An additional effect that can be observed is the time-dependent change of the bending of the structure. This effect is called backrelaxation. In the current project, we applied electrochemical reactions at the electrodes to the electro-chemical multi-field formulation (Poisson-Nernst-Planck formalism). A time-dependent relaxation effect was observed from the numerical simulation outputs. However, the approach with electrochemical reactions did not match to the experimental results, so that the postulate could be ruled out as the sole cause for backrelaxation in IPMC.
This research paper was produced in course of my studies. It was conducted at TU Dresden and supervised by Prof. Thomas Wallmersperger (Chair of Mechanics of Multifunctional Structures). I kindly thank for his support.
Dual-phase steel consists of two different components with ferritic and martensitic structure. On the level of technical components, the microstructure can be neglected by means of homogenisation in Finite Element simulations. A first step in this homogenisation is the identification of grain properties in the microstructure. To identify grain properties from two-dimensional micrographs, a Matlab script for image processing was developed. With this script, the engineer interested in homogenisation of dual phase steel is capable of identifying properties like circumference, in-circle and orientation of all corns in the microstructure.
The research was conducted at TU Dresden and supervised by Dr.-Ing. Martin Hofmann. I am very grateful for his support.
While in experimental mechanics, the evaluation of measurement errors and their propagation besides the errors of the measurement principle are thoroughly regarded, the error evaluation in computational mechanics is mostly limited to convergence observation for numerical schemes. Nevertheless, hardware-inflicted errors can be crucial to the result of a calculation that seems to be much more accurate than the experimental equivalent measure. Bad experience with the usage of the IEEE 754 standard single precision lead for many numericists to statements like 'I always use double precision, it feels safer!'. Regarding the inherent difference in performance and memory usage, the scientist needs to verify, which precision (i.e. which choice of IEEE floating point representation data type) is really needed for a calculation. It is therefore reasonable to apply proved tools of numerical accuracy, e.g. the here treated CADNA-library or the PRECISE-toolbox, to computational mechanics, represented by computational fluid mechanics.
This report aims to discover the influences of the computational error on CFD results in direct numerical simulation (DNS). It is strongly linked to the industrial placement report which was created in parallel to show the practical aspects of the work, which is described theoretically herein.
The research was carried out and the project report was written under supervision of and in cooperation with Thiên-Hiêp Lê at DSNA/ONERA - The French Aerospace Lab, Fabienne Jézéquel at PEQUAN/LIP6 - Laboratoire d'Informatique de Paris 6 and Prof. Jochen Fröhlich at TU Dresden (Chair of Fluid Mechanics). I kindly thank all partners for their support.