Browsing by Author "Röhm, Dominic"

Now showing 1 - 1 of 1

Open Access
Multiscale simulations of soft and hard matter
(2015) Röhm, Dominic; Holm, Christian (Prof. Dr.)
The first part of this Thesis presents results of investigations of selected dynamical properties. Furthermore, die crystallization of colloidal suspensions has been investigated. We show that for charge-stabilized suspensions, where the colloids interact via the Yukawa potential, hydrodynamic interactions can have a remarkable impact on the crystallization of colloidal particles. The results are based on Molecular Dynamics (MD) simulations of heterogeneous crystallization in a suspension of charged colloids supported by the computation of the solvent dynamics by the Lattice-Boltzmann (LB) method. In order to investigate the role of hydrodynamic interactions mediated by the solvent, we modeled the solvent both implicitly and explicitly, using Langevin dynamics and the fluctuating LB method, respectively. Our simulations show a reduction of the crystal growth velocity due to hydrodynamic interactions even at moderate hydrodynamic coupling. The slow down of the crystallization is accompanied by narrowing of the pre-ordering region, which shows that the attachment to a crystal surface is not a purely long-time diffusive process, as commonly thought. The arrangement of the colloids in the early state of a new crystal layer seems to be affected by the short-time dynamics of the colloids, which is again affected by hydrodynamic interactions. Crystallization in suspensions therefore can differ strongly from that of pure melts. In the second part of this Thesis we will introduce an approach for the efficient computation of strain evolution in a copper crystal. Here, instead of attaching a continuum solver to an MD simulation, we used a method that combines a finite-volume solver and MD simulations by spawning independent MD simulations to include microscopic details into the stress computation, which serves as input for every finite volume at the macro level. We developed an adaptive sampling method called Distributed Database Kriging for Adaptive Sampling, which applies a prediction scheme known as kriging to the heterogeneous multiscale method (HMM) for stochastic data supported by a cloud database. We demonstrated by means of two elastodynamics test problems, that a speedup of a factor of 2.5 to 25 can be achieved.