Ruslan Puscasu | The University of Queensland, Australia (original) (raw)

Papers by Ruslan Puscasu

This study aims to demonstrate the capacity of the Discrete Element Method (DEM) to model and sim... more This study aims to demonstrate the capacity of the Discrete Element Method (DEM) to model and simulate fracture in brittle rock, capture the crack propagation process and determine crack propagation rates for various levels of external loading. In the numerical experiments reported here, a macroscopically homogeneous body of brittle rock, containing a single pre-existing crack, was subjected to tensile stress loading. Simulations concerned with model I crack growth process in 3D were carried out. The process of fracture has been observed under various steady state and dynamic conditions. The numerical results obtained reproduce qualitatively and quantitatively a variety of observed fracture phenomena. Specifically, this study examined the onset of cracks, single dominant cracks that grow across the specimen, crack bifurcations and branching. The density of fractures that can be nucleated and grow to a significant size with and without arresting each other, crack growth speed and stress distribution, all have been consistently and systematically analyzed. Suggestions on potential future developments are made and discussed.

Because of its complexity, the 3D Ising model has not been given an exact analytic solution so fa... more Because of its complexity, the 3D Ising model has not been given an exact analytic solution so far, as well as the 2D Ising in non zero external field conditions. In real materials the phase transition creates a discontinuity. We analysed the Ising model that presents similar discontinuities. We use Monte Carlo methods with a single spin change or a spin cluster change to calculate macroscopic quantities, such as specific heat and magnetic susceptibility. We studied the differences between these methods. Local MC algorithms (such as Metropolis) perform poorly for large lattices because they update only one spin at a time, so it takes many iterations to get a statistically independent configuration. More recent spin cluster algorithms use clever ways of finding clusters of sites that can be updated at once. The single cluster method is probably the best sequential cluster algorithm. We also used the entropic sampling method to simulate the density of states. This method takes into account all possible configurations, not only the most probable. The entropic method also gives good results in the 3D case. We studied the usefulness of distributed computing for Ising model. We established a parallelization strategy to explore Metropolis Monte Carlo simulation and Swendsen-Wang Monte Carlo simulation of this spin model using the data parallel languages on different platform. After building a computer cluster we made a Monte Carlo estimation of 2D and 3D Ising thermodynamic properties and compare the results with the sequential computing. In the same time we made quantitative analysis such as speed up and efficiency for different sets of combined parameters (e.g. lattice size, parallel algorithms, chosen model).

Procedia Computer Science, 2014

In this paper, an implementation study was undertaken to employ Artificial Neural Networks (ANN) ... more In this paper, an implementation study was undertaken to employ Artificial Neural Networks (ANN) in third-generation ocean wave models for direct mapping of wind-wave spectra into exact nonlinear interactions. While the investigation expands on previously reported feasibility studies of Neural Network Interaction Approximations (NNIA), it focuses on a new robust neural network that is implemented in Wavewatch III (WW3) model. Several idealistic and real test scenarios were carried out. The obtained results confirm the feasibility of NNIA in terms of speeding-up model calculations and is fully capable of providing operationally acceptable model integrations. The ANN is able to emulate the exact nonlinear interaction for single-and multimodal wave spectra with a much higher accuracy then Discrete Interaction Approximation (DIA). NNIA performs at least twice as fast as DIA and at least two hundred times faster than exact method (Web-Resio-Tracy, WRT) for a well trained dataset. The accuracy of NNIA is network configuration dependent. For most optimal network configurations, the NNIA results and scatter statistics show good agreement with exact results by means of growth curves and integral parameters. Practical possibilities for further improvements in achieving fast and highly accurate emulations using ANN for emulating time consuming exact nonlinear interactions are also suggested and discussed.

Physical Review E, 2010

The wave-vector dependent shear viscosities for butane and freely jointed chains have been determ... more The wave-vector dependent shear viscosities for butane and freely jointed chains have been determined. The transverse momentum density and stress autocorrelation functions have been determined by equilibrium molecular dynamics in both atomic and molecular hydrodynamic formalisms. The density, temperature, and chain length dependencies of the reciprocal and real-space viscosity kernels are presented. We find that the density has a major effect on the shape of the kernel. The temperature range and chain lengths considered here have by contrast less impact on the overall normalized shape. Functional forms that fit the wave-vector-dependent kernel data over a large density and wave-vector range have also been tested. Finally, a structural normalization of the kernels in physical space is considered. Overall, the real-space viscosity kernel has a width of roughly 3-6 atomic diameters, which means that generalized hydrodynamics must be applied in predicting the flow properties of molecular fluids on length scales where the strain rate varies sufficiently in the order of these dimensions ͑e.g., nanofluidic flows͒.

Journal of Physics: Condensed Matter, 2010

We present an extended analysis of the wavevector dependent shear viscosity of monatomic and diat... more We present an extended analysis of the wavevector dependent shear viscosity of monatomic and diatomic (liquid chlorine) fluids over a wide range of wavevectors and for a variety of state points. The analysis is based on equilibrium molecular dynamics simulations, which involve the evaluation of transverse momentum density and shear stress autocorrelation functions. For liquid chlorine we present the results in both atomic and molecular formalisms. We find that the viscosity kernel of chlorine in the atomic representation is statistically indistinguishable from that in the molecular representation. The results further suggest that the real space viscosity kernels of monatomic and diatomic fluids depend sensitively on the density, the potential energy function and the choice of fitting function in reciprocal space. It is also shown that the reciprocal space shear viscosity data can be fitted to two different simple functional forms over the entire density, temperature and wavevector range: a function composed of n-Gaussian terms and a Lorentzian-type function. Overall, the real space viscosity kernel has a width of 3-6 atomic diameters, which means that the generalized hydrodynamic constitutive relation is required for fluids with strain rates that vary nonlinearly over distances of the order of atomic dimensions.

The Journal of Chemical Physics, 2010

The nonlocal viscosity kernels of polymer melts have been determined by means of equilibrium mole... more The nonlocal viscosity kernels of polymer melts have been determined by means of equilibrium molecular dynamics upon cooling toward the glass transition. Previous results for the temperature dependence of the self-diffusion coefficient and the value of the glass transition temperature are confirmed. We find that it is essential to include the attractive part of the interatomic potential in order to observe a strong glass transition. The width of the reciprocal space kernel decreases dramatically near the glass transition, being described by a deltalike function near and below the glass transition, leading to a very broad kernel in physical space. Thus, spatial nonlocality turns out to play an important role in polymeric fluids at temperatures near the glass transition temperature.

Acta Physica Slovaca. Reviews and Tutorials, 2000

The paper gives a review of recent advances in theory and simulation of nonlocal transport in nan... more The paper gives a review of recent advances in theory and simulation of nonlocal transport in nanoflows. The aim is to show how to computationally model and simulate the nonlocal viscous transport in atomic and molecular fluids. The ultimate goal is to provide nanofluidics and other disciplines with methodologies capable to give exact descriptions of flow at the nanoscale by using nonlocal constitutive relations which involve nonlocal transport kernels.

This study aims to demonstrate the capacity of the Discrete Element Method (DEM) to model and sim... more This study aims to demonstrate the capacity of the Discrete Element Method (DEM) to model and simulate fracture in brittle rock, capture the crack propagation process and determine crack propagation rates for various levels of external loading. In the numerical experiments reported here, a macroscopically homogeneous body of brittle rock, containing a single pre-existing crack, was subjected to tensile stress loading. Simulations concerned with model I crack growth process in 3D were carried out. The process of fracture has been observed under various steady state and dynamic conditions. The numerical results obtained reproduce qualitatively and quantitatively a variety of observed fracture phenomena. Specifically, this study examined the onset of cracks, single dominant cracks that grow across the specimen, crack bifurcations and branching. The density of fractures that can be nucleated and grow to a significant size with and without arresting each other, crack growth speed and stress distribution, all have been consistently and systematically analyzed. Suggestions on potential future developments are made and discussed.

Because of its complexity, the 3D Ising model has not been given an exact analytic solution so fa... more Because of its complexity, the 3D Ising model has not been given an exact analytic solution so far, as well as the 2D Ising in non zero external field conditions. In real materials the phase transition creates a discontinuity. We analysed the Ising model that presents similar discontinuities. We use Monte Carlo methods with a single spin change or a spin cluster change to calculate macroscopic quantities, such as specific heat and magnetic susceptibility. We studied the differences between these methods. Local MC algorithms (such as Metropolis) perform poorly for large lattices because they update only one spin at a time, so it takes many iterations to get a statistically independent configuration. More recent spin cluster algorithms use clever ways of finding clusters of sites that can be updated at once. The single cluster method is probably the best sequential cluster algorithm. We also used the entropic sampling method to simulate the density of states. This method takes into account all possible configurations, not only the most probable. The entropic method also gives good results in the 3D case. We studied the usefulness of distributed computing for Ising model. We established a parallelization strategy to explore Metropolis Monte Carlo simulation and Swendsen-Wang Monte Carlo simulation of this spin model using the data parallel languages on different platform. After building a computer cluster we made a Monte Carlo estimation of 2D and 3D Ising thermodynamic properties and compare the results with the sequential computing. In the same time we made quantitative analysis such as speed up and efficiency for different sets of combined parameters (e.g. lattice size, parallel algorithms, chosen model).

Procedia Computer Science, 2014

In this paper, an implementation study was undertaken to employ Artificial Neural Networks (ANN) ... more In this paper, an implementation study was undertaken to employ Artificial Neural Networks (ANN) in third-generation ocean wave models for direct mapping of wind-wave spectra into exact nonlinear interactions. While the investigation expands on previously reported feasibility studies of Neural Network Interaction Approximations (NNIA), it focuses on a new robust neural network that is implemented in Wavewatch III (WW3) model. Several idealistic and real test scenarios were carried out. The obtained results confirm the feasibility of NNIA in terms of speeding-up model calculations and is fully capable of providing operationally acceptable model integrations. The ANN is able to emulate the exact nonlinear interaction for single-and multimodal wave spectra with a much higher accuracy then Discrete Interaction Approximation (DIA). NNIA performs at least twice as fast as DIA and at least two hundred times faster than exact method (Web-Resio-Tracy, WRT) for a well trained dataset. The accuracy of NNIA is network configuration dependent. For most optimal network configurations, the NNIA results and scatter statistics show good agreement with exact results by means of growth curves and integral parameters. Practical possibilities for further improvements in achieving fast and highly accurate emulations using ANN for emulating time consuming exact nonlinear interactions are also suggested and discussed.

Physical Review E, 2010

The wave-vector dependent shear viscosities for butane and freely jointed chains have been determ... more The wave-vector dependent shear viscosities for butane and freely jointed chains have been determined. The transverse momentum density and stress autocorrelation functions have been determined by equilibrium molecular dynamics in both atomic and molecular hydrodynamic formalisms. The density, temperature, and chain length dependencies of the reciprocal and real-space viscosity kernels are presented. We find that the density has a major effect on the shape of the kernel. The temperature range and chain lengths considered here have by contrast less impact on the overall normalized shape. Functional forms that fit the wave-vector-dependent kernel data over a large density and wave-vector range have also been tested. Finally, a structural normalization of the kernels in physical space is considered. Overall, the real-space viscosity kernel has a width of roughly 3-6 atomic diameters, which means that generalized hydrodynamics must be applied in predicting the flow properties of molecular fluids on length scales where the strain rate varies sufficiently in the order of these dimensions ͑e.g., nanofluidic flows͒.

Journal of Physics: Condensed Matter, 2010

We present an extended analysis of the wavevector dependent shear viscosity of monatomic and diat... more We present an extended analysis of the wavevector dependent shear viscosity of monatomic and diatomic (liquid chlorine) fluids over a wide range of wavevectors and for a variety of state points. The analysis is based on equilibrium molecular dynamics simulations, which involve the evaluation of transverse momentum density and shear stress autocorrelation functions. For liquid chlorine we present the results in both atomic and molecular formalisms. We find that the viscosity kernel of chlorine in the atomic representation is statistically indistinguishable from that in the molecular representation. The results further suggest that the real space viscosity kernels of monatomic and diatomic fluids depend sensitively on the density, the potential energy function and the choice of fitting function in reciprocal space. It is also shown that the reciprocal space shear viscosity data can be fitted to two different simple functional forms over the entire density, temperature and wavevector range: a function composed of n-Gaussian terms and a Lorentzian-type function. Overall, the real space viscosity kernel has a width of 3-6 atomic diameters, which means that the generalized hydrodynamic constitutive relation is required for fluids with strain rates that vary nonlinearly over distances of the order of atomic dimensions.

The Journal of Chemical Physics, 2010

The nonlocal viscosity kernels of polymer melts have been determined by means of equilibrium mole... more The nonlocal viscosity kernels of polymer melts have been determined by means of equilibrium molecular dynamics upon cooling toward the glass transition. Previous results for the temperature dependence of the self-diffusion coefficient and the value of the glass transition temperature are confirmed. We find that it is essential to include the attractive part of the interatomic potential in order to observe a strong glass transition. The width of the reciprocal space kernel decreases dramatically near the glass transition, being described by a deltalike function near and below the glass transition, leading to a very broad kernel in physical space. Thus, spatial nonlocality turns out to play an important role in polymeric fluids at temperatures near the glass transition temperature.

Acta Physica Slovaca. Reviews and Tutorials, 2000

The paper gives a review of recent advances in theory and simulation of nonlocal transport in nan... more The paper gives a review of recent advances in theory and simulation of nonlocal transport in nanoflows. The aim is to show how to computationally model and simulate the nonlocal viscous transport in atomic and molecular fluids. The ultimate goal is to provide nanofluidics and other disciplines with methodologies capable to give exact descriptions of flow at the nanoscale by using nonlocal constitutive relations which involve nonlocal transport kernels.