Frank Raischel

Quantitative analysis of numerical estimates for the permeability of porous media from lattice-Boltzmann simulations

During the last decade, lattice-Boltzmann simulations have been improved to become an efficient tool for determining the permeability of porous media samples. However, well-known improvements of the original algorithm are often not implemented. These include, for example, multirelaxation time schemes or improved boundary conditions, as well as different possibilities to impose a pressure gradient. This paper shows that a significant difference of the calculated permeabilities can be found unless one uses a carefully selected set-up. We present a detailed discussion of possible simulation set-ups and quantitative studies of the influence of simulation parameters. We illustrate our results by applying the algorithm to a Fontainebleau sandstone and by comparing our benchmark studies to other numerical permeability measurements in the literature.

Air quality prediction using optimal neural networks with stochastic variables

We apply recent methods in stochastic data analysis for discovering a set of few stochastic variables that represent the relevant information on a multivariate stochastic system, used as input for artificial neural network models for air quality forecast. We show that using these derived variables as input variables for training the neural networks it is possible to significantly reduce the amount of input variables necessary for the neural network model, without considerably changing the predictive power of the model. The reduced set of variables including these derived variables is therefore proposed as an optimal variable set for training neural network models in forecasting geophysical and weather properties. Finally, we briefly discuss other possible applications of such optimized neural network models.

Dose properties of a laser accelerated electron beam and prospects for clinical application.

Laser wakefield acceleration (LWFA) technology has evolved to where it should be evaluated for its potential as a future competitor to existing technology that produces electron and x-ray beams. The purpose of the present work is to investigate the dosimetric properties of an electron beam that should be achievable using existing LWFA technology, and to document the necessary improvements to make radiotherapy application for LWFA viable. This paper first qualitatively reviews the fundamental principles of LWFA and describes a potential design for a 30 cm accelerator chamber containing a gas target. Electron beam energy spectra, upon which our dose calculations are based, were obtained from a uniform energy distribution and from two-dimensional particle-in-cell (2D PIC) simulations. The 2D PIC simulation parameters are consistent with those reported by a previous LWFA experiment. According to the 2D PIC simulations, only approximately 0.3% of the LWFA electrons are emitted with an energy greater than 1 MeV. We studied only the high-energy electrons to determine their potential for clinical electron beams of central energy from 9 to 21 MeV. Each electron beam was broadened and flattened by designing a dual scattering foil system to produce a uniform beam (103%>off-axis ratio>95%) over a 25 x 25 cm2 field. An energy window (deltaE) ranging from 0.5 to 6.5 MeV was selected to study central-axis depth dose, beam flatness, and dose rate. Dose was calculated in water at a 100 cm source-to-surface distance using the EGS/BEAM Monte Carlo algorithm. Calculations showed that the beam flatness was fairly insensitive to deltaE. However, since the falloff of the depth-dose curve (R10-R90) and the dose rate both increase with deltaE, a tradeoff between minimizing (R10-R90) and maximizing dose rate is implied. If deltaE is constrained so that R10-R90 is within 0.5 cm of its value for a monoenergetic beam, the maximum practical dose rate based on 2D PIC is approximately 0.1 Gy min(-1) for a 9 MeV beam and 0.03 Gy min(-1) for a 15 MeV beam. It was concluded that current LWFA technology should allow a table-top terawatt (T3) laser to produce therapeutic electron beams that have acceptable flatness, penetration, and falloff of depth dose; however, the dose rate is still 1%-3% of that which would be acceptable, especially for higher-energy electron beams. Further progress in laser technology, e.g., increasing the pulse repetition rate or number of high energy electrons generated per pulse, is necessary to give dose rates acceptable for electron beams. Future measurements confirming dosimetric calculations are required to substantiate our results. In addition to achieving adequate dose rate, significant engineering developments are needed for this technology to compete with current electron acceleration technology. Also, the functional benefits of LWFA electron beams require further study and evaluation.

Failure process of a bundle of plastic fibers

We present an extension of fiber bundle models considering that failed fibers still carry a fraction 0 < or = alpha < or = 1 of their failure load. The value of alpha interpolates between the perfectly brittle failure (alpha = 0) and perfectly plastic behavior (alpha = 1) of fibers. We show that the finite load bearing capacity of broken fibers has a substantial effect on the failure process of the bundle. In the case of global load sharing it is found that for alpha --> 1 the macroscopic response of the bundle becomes perfectly plastic with a yield stress equal to the average fiber strength. On the microlevel, the size distribution of avalanches has a crossover from a power law of exponent approximately 2.5 to a faster exponential decay. For localized load sharing, computer simulations revealed a sharp transition at a well-defined value alpha(c) from a phase where macroscopic failure occurs due to localization as a consequence of local stress enhancements, to another one where the disordered fiber strength dominates the damage process. Analyzing the microstructure of damage, the transition proved to be analogous to percolation. At the critical point alpha(c), the spanning cluster of damage is found to be compact with a fractal boundary. The distribution of bursts of fiber breakings shows a power-law behavior with a universal exponent approximately 1.5 equal to the mean-field exponent of fiber bundles of critical strength distributions. The model can be relevant to understand the shear failure of glued interfaces where failed regions can still transmit load by remaining in contact.

Local load sharing fiber bundles with a lower cutoff of strength disorder

We study the failure properties of fiber bundles with a finite lower cutoff of the strength disorder varying the range of interaction between the limiting cases of completely global and completely local load sharing. Computer simulations revealed that at any range of load redistribution there exists a critical cutoff strength where the macroscopic response of the bundle becomes perfectly brittle, i.e., linearly elastic behavior is obtained up to global failure, which occurs catastrophically after the breaking of a small number of fibers. As an extension of recent mean field studies [Phys. Rev. Lett. 95, 125501 (2005)], we demonstrate that approaching the critical cutoff, the size distribution of bursts of breaking fibers shows a crossover to a universal power law form with an exponent 3/2 independent of the range of interaction.

Continuous Damage Fiber Bundle Model for Strongly Disordered Materials

We present an extension of the continuous damage fiber bundle model to describe the gradual degradation of highly heterogeneous materials under an increasing external load. The breaking of a fiber in the model is preceded by a sequence of partial failure events occurring at random threshold values. In order to capture the subsequent propagation and arrest of cracks, furthermore, the disorder of the number of degradation steps of material constituents, the failure thresholds of single fibers, are sorted into ascending order and their total number is a Poissonian distributed random variable over the fibers. Analytical and numerical calculations showed that the failure process of the system is governed by extreme value statistics, which has a substantial effect on the macroscopic constitutive behavior and on the microscopic bursting activity as well.

Simple beam model for the shear failure of interfaces

We propose a model for the shear failure of a glued interface between two solid blocks. We model the interface as an array of elastic beams which experience stretching and bending under shear load and break if the two deformation modes exceed randomly distributed breaking thresholds. The two breaking modes can be independent or combined in the form of a von Mises-type breaking criterion. Assuming global load sharing following the beam breaking, we obtain analytically the macroscopic constitutive behavior of the system and describe the microscopic process of the progressive failure of the interface. We work out an efficient simulation technique which allows for the study of large systems. The limiting case of very localized interaction of surface elements is explored by computer simulations.

Neural network forecast of daily pollution concentration using optimal meteorological data at synoptic and local scales

We present a simple neural network and data pre–selection framework, discriminating the most essential input data for accurately forecasting the concentrations of PM10, based on observations for the years between 2002 and 2006 in the metropolitan region of Lisbon, Portugal. Starting from a broad panoply of different data sets collected at several air quality and meteorological stations, a forward stepwise regression procedure is applied enabling to automatically identify the most important variables for predicting the pollutant and also to rank them in order of importance. The importance of this variable ranking is discussed, showing that it is very sensitive to the urban location where measurements are obtained. Additionally, the importance of Circulation Weather Types is highlighted, characterizing synoptic scale circulation patterns and the concentration of pollutants. We then quantify the performance of linear and non–linear neural network models when applied to PM10 concentrations. In the light of contradictory results of previous studies, our results show no clear superiority for the case studied of non–linear models over linear models. While all models show similar predictive performances, we find important differences in false alarm rates and demonstrate the importance of removing weekly cycles from input variables.

Uncovering wind turbine properties through two-dimensional stochastic modeling of wind dynamics

Using a method for stochastic data analysis borrowed from statistical physics, we analyze synthetic data from a Markov chain model that reproduces measurements of wind speed and power production in a wind park in Portugal. We show that our analysis retrieves indeed the power performance curve, which yields the relationship between wind speed and power production, and we discuss how this procedure can be extended for extracting unknown functional relationships between pairs of physical variables in general. We also show how specific features, such as the rated speed of the wind turbine or the descriptive wind speed statistics, can be related to the equations describing the evolution of power production and wind speed at single wind turbines.

Principal axes for stochastic dynamics

We introduce a general procedure for directly ascertaining how many independent stochastic sources exist in a complex system modeled through a set of coupled Langevin equations of arbitrary dimension. The procedure is based on the computation of the eigenvalues and the corresponding eigenvectors of local diffusion matrices. We demonstrate our algorithm by applying it to two examples of systems showing Hopf bifurcation. We argue that computing the eigenvectors associated to the eigenvalues of the diffusion matrix at local mesh points in the phase space enables one to define vector fields of stochastic eigendirections. In particular, the eigenvector associated to the lowest eigenvalue defines the path of minimum stochastic forcing in phase space, and a transform to a new coordinate system aligned with the eigenvectors can increase the predictability of the system.