Simulación eficiente de mecanismos flexibles mediante formulaciones topológicas

El objetivo de esta Tesis es presentar un método eficiente para la evaluación de sistemas multi-cuerpo con elementos flexibles con pequeñas deformaciones, basado en métodos topológicos para la simulación de sistemas tan complejos como los que se utilizan en la práctica y en tiempo real o próximo al real. Se ha puesto un especial énfasis en la resolución eficiente de aquellos aspectos que conllevan mayor coste computacional, tales como la evaluación de las ecuaciones dinámicas y el cálculo de los términos de inercia. Las ecuaciones dinámicas se establecen en función de las variables independientes del sistema, y la integración de las mismas se realiza mediante formulaciones implícitas de index-3. Esta Tesis se articula en seis Capítulos. En el Capítulo 1 se realiza una revisión bibliográfica de la simulación de sistemas flexibles y los métodos más relevantes de integración de las ecuaciones diferenciales del movimiento. Asimismo, se presentan los objetivos de esta Tesis. En el Capítulo 2 se presenta un método semi-recursivo para la evaluación de las ecuaciones de los sistemas multi-cuerpo con elementos flexibles basado en formulaciones topológicas y síntesis modal. Esta Tesis determina la posición de cada punto del cuerpo flexible en función de un sistema de referencia flotante que se mueve con dicho cuerpo y de las amplitudes de ciertos modos de deformación calculados a partir de un mallado obtenido mediante el Método de Elementos Finitos. Se presta especial atención en las condiciones de contorno que se han de tener en cuenta a la hora de establecer las variables que definen la deformación del cuerpo flexible. El Capítulo 3 se centra en la evaluación de los términos de inercia de los sistemas flexibles que generalmente conllevan un alto coste computacional. Se presenta un método que permite el cálculo de dichos términos basado en el uso de 24 matrices constantes que pueden ser calculadas previamente al proceso de integración. Estas matrices permiten evaluar la matriz de masas y el vector de fuerzas de inercia dependientes de la velocidad sin que sea necesario evaluar la posición deformada de todos los puntos del cuerpo flexible. Se realiza un análisis pormenorizado de dichas matrices con el objetivo de optimizar su cálculo estableciendo aproximaciones que permitan reducir el número de dichos términos y optimizar aún más su evaluación. Se analizan dos posibles simplificaciones: la primera utiliza una discretización no-consistente basada en elementos finitos en los que se definen únicamente los desplazamientos axiales de los nodos; en la segunda propuesta se hace uso de una matriz de masas concentradas (Lumped Mass). Basándose en la formulación presentada, el Capítulo 4 aborda la integración eficiente de las ecuaciones dinámicas. Se presenta un método iterativo para la integración con fórmulas de index-3 basado en la proyección de las ecuaciones dinámicas según las variables independientes del sistema multi-cuerpo. El cálculo del residuo del sistema de ecuaciones no lineales que se ha de resolver de modo iterativo se realiza mediante un proceso recursivo muy eficiente que aprovecha la estructura topológica del sistema. Se analizan tres formas de evaluar la matriz tangente del citado sistema no lineal: evaluación aproximada, numérica y recursiva. El método de integración presentado permite el uso de distintas fórmulas. En esta Tesis se analizan la Regla Trapezoidal, la fórmula BDF de segundo orden y un método híbrido TR-BDF2. Para este último caso se presenta un algoritmo de paso variable. En el Capítulo 5 plantea la implementación del método propuesto en un programa general de simulación de mecanismos que permita la resolución de cualquier sistema multi-cuerpo definiéndolo mediante un fichero de datos. La implementación de este programa se ha realizado tanto en C++ como en Java. Se muestran los resultados de las formulaciones presentadas en esta Tesis mediante la simulación de cuatro ejemplos de distinta complejidad. Mediante análisis concretos se comparan la formulación presentada con otras existentes. También se analiza el efecto del lenguaje de programación utilizado en la implementación y los efectos de las posibles simplificaciones planteadas. Por último, el Capítulo 6 resume las principales conclusiones alcanzadas en la Tesis y las futuras líneas de investigación que con ella se abren. ABSTRACT This Thesis presents an efficient method for solving the forward dynamics of a multi-body sys-tem formed by rigid and flexible bodies with small strains for real-time simulation of real-life models. It is based on topological formulations. The presented work focuses on the efficient solution of the most time-consuming tasks of the simulation process, such as the numerical integration of the motion differential equations and in particular the evaluation of the inertia terms corresponding to the flexible bodies. The dynamic equations are formulated in terms of independent variables of the muti-body system, and they are integrated by means of implicit index-3 formulae. The Thesis is arranged in six chapters. Chapter 1 presents a review of the most relevant and recent contributions related to the modelization of flexible multi-body systems and the integration of the corresponding dynamic equations. The main objectives of the Thesis are also presented in detail. Chapter 2 presents a semi-recursive method for solving the equations of a multi-body system with flexible bodies based on topological formulations and modal synthesis. This Thesis uses the floating frame approach and the modal amplitudes to define the position of any point at the flexible body. These modal deformed shapes are obtained by means of the Finite Element Method. Particular attention has been taken to the boundary conditions used to define the deformation of the flexible bodies. Chapter 3 focuses on the evaluation of the inertia terms, which is usually a very time-consuming task. A new method based on the use of 24 constant matrices is presented. These matrices are evaluated during the set-up step, before the integration process. They allow the calculation of the inertia terms in terms of the position and orientation of the local coordinate system and the deformation variables, and there is no need to evaluate the position and velocities of all the nodes of the FEM mesh. A deep analysis of the inertia terms is performed in order to optimize the evaluation process, reducing both the terms used and the number of arithmetic operations. Two possible simplifications are presented: the first one uses a non-consistent approach in order to define the inertia terms respect to the Cartesian coordinates of the FEM mesh, rejecting those corresponding to the angular rotations; the second approach makes use of lumped mass matrices. Based on the previously presented formulation, Chapter 4 is focused on the numerical integration of the motion differential equations. A new predictor-corrector method based on index-3 formulae and on the use of multi-body independent variables is presented. The evaluation of the dynamic equations in a new time step needs the solution of a set on nonlinear equations by a Newton-Raphson iterative process. The computation of the corresponding residual vector is performed efficiently by taking advantage of the system’s topological structure. Three methods to compute the tangent matrix are presented: an approximated evaluation that considers only the most relevant terms, a numerical approach based on finite differences and a recursive method that uses the topological structure. The method presented for integrating the dynamic equations can use a variety of integration formulae. This Thesis analyses the use of the trapezoidal rule, the 2nd order BDF formula and the hybrid TR-BDF2 method. A variable-time step strategy is presented for the last one. Chapter 5 describes the implementation of the proposed method in a general purpose pro-gram for solving any multibody defined by a data file. This program is implemented both in C++ and Java. Four examples are used to check the validity of the formulation and to compare this method with other methods commonly used to solve the dynamic equations of multi-body systems containing flexible bodies. The efficiency of the programming methodology used and the effect of the possible simplifications proposed are also analyzed. Chapter 6 summarizes the main Conclusions obtained in this Thesis and the new lines of research that have been opened.

On ANOVA Decompositions of Kernels and Gaussian Random Field Paths

The FANOVA (or “Sobol’-Hoeffding”) decomposition of multivariate functions has been used for high-dimensional model representation and global sensitivity analysis. When the objective function f has no simple analytic form and is costly to evaluate, computing FANOVA terms may be unaffordable due to numerical integration costs. Several approximate approaches relying on Gaussian random field (GRF) models have been proposed to alleviate these costs, where f is substituted by a (kriging) predictor or by conditional simulations. Here we focus on FANOVA decompositions of GRF sample paths, and we notably introduce an associated kernel decomposition into 4 d 4d terms called KANOVA. An interpretation in terms of tensor product projections is obtained, and it is shown that projected kernels control both the sparsity of GRF sample paths and the dependence structure between FANOVA effects. Applications on simulated data show the relevance of the approach for designing new classes of covariance kernels dedicated to high-dimensional kriging.

Automatic multirate methods for ordinary differential equations /

"UILU-ENG 80 1701."

Optimal stiffly stable methods for ordinary differential equations,

"COO-1469-0164."

Comparison of Lagrangian Coherent Structures and Relative Dispersion in a Mixing Layer

Thesis (Master's)--University of Washington, 2016-06

Nonlinear dynamics of a satellite with deployable solar panel arrays

The multibody dynamics of a satellite in circular orbit, modeled as a central body with two hinge-connected deployable solar panel arrays, is investigated. Typically, the solar panel arrays are deployed in orbit using preloaded torsional springs at the hinges in a near symmetrical accordion manner, to minimize the shock loads at the hinges. There are five degrees of freedom of the interconnected rigid bodies, composed of coupled attitude motions (pitch, yaw and roll) of the central body plus relative rotations of the solar panel arrays. The dynamical equations of motion of the satellite system are derived using Kane's equations. These are then used to investigate the dynamic behavior of the system during solar panel deployment via the 7-8th-order Runge-Kutta integration algorithms and results are compared with approximate analytical solutions. Chaotic attitude motions of the completely deployed satellite in circular orbit under the influence of the gravity-gradient torques are subsequently investigated analytically using Melnikov's method and confirmed via numerical integration. The Hamiltonian equations in terms of Deprit's variables are used to facilitate the analysis. (C) 2003 Published by Elsevier Ltd.

Photodissociation of benzene under collision-free conditions: An ab initio/Rice-Ramsperger-Kassel-Marcus study

The ab initio/Rice-Ramsperger-Kassel-Marcus (RRKM) approach has been applied to investigate the photodissociation mechanism of benzene at various wavelengths upon absorption of one or two UV photons followed by internal conversion into the ground electronic state. Reaction pathways leading to various decomposition products have been mapped out at the G2M level and then the RRKM and microcanonical variational transition state theories have been applied to compute rate constants for individual reaction steps. Relative product yields (branching ratios) for C6H5+H, C6H4+H-2, C4H4+C2H2, C4H2+C2H4, C3H3+C3H3, C5H3+CH3, and C4H3+C2H3 have been calculated subsequently using both numerical integration of kinetic master equations and the steady-state approach. The results show that upon absorption of a 248 nm photon dissociation is too slow to be observable in molecular beam experiments. In photodissociation at 193 nm, the dominant dissociation channel is H atom elimination (99.6%) and the minor reaction channel is H-2 elimination, with the branching ratio of only 0.4%. The calculated lifetime of benzene at 193 nm is about 11 mus, in excellent agreement with the experimental value of 10 mus. At 157 nm, the H loss remains the dominant channel but its branching ratio decreases to 97.5%, while that for H-2 elimination increases to 2.1%. The other channels leading to C3H3+C3H3, C5H3+CH3, C4H4+C2H2, and C4H3+C2H3 play insignificant role but might be observed. For photodissociation upon absorption of two UV photons occurring through the neutral hot benzene mechanism excluding dissociative ionization, we predict that the C6H5+H channel should be less dominant, while the contribution of C6H4+H-2 and the C3H3+C3H3, CH3+C5H3, and C4H3+C2H3 radical channels should significantly increase. (C) 2004 American Institute of Physics.

Methods for Categorical Longitudinal Survey Data: Understanding Employment Status of Australian Women

Many variables that are of interest in social science research are nominal variables with two or more categories, such as employment status, occupation, political preference, or self-reported health status. With longitudinal survey data it is possible to analyse the transitions of individuals between different employment states or occupations (for example). In the statistical literature, models for analysing categorical dependent variables with repeated observations belong to the family of models known as generalized linear mixed models (GLMMs). The specific GLMM for a dependent variable with three or more categories is the multinomial logit random effects model. For these models, the marginal distribution of the response does not have a closed form solution and hence numerical integration must be used to obtain maximum likelihood estimates for the model parameters. Techniques for implementing the numerical integration are available but are computationally intensive requiring a large amount of computer processing time that increases with the number of clusters (or individuals) in the data and are not always readily accessible to the practitioner in standard software. For the purposes of analysing categorical response data from a longitudinal social survey, there is clearly a need to evaluate the existing procedures for estimating multinomial logit random effects model in terms of accuracy, efficiency and computing time. The computational time will have significant implications as to the preferred approach by researchers. In this paper we evaluate statistical software procedures that utilise adaptive Gaussian quadrature and MCMC methods, with specific application to modeling employment status of women using a GLMM, over three waves of the HILDA survey.

A study of the micro-mechanics of granular media

The development of more realistic constitutive models for granular media, such as sand, requires ingredients which take into account the internal micro-mechanical response to deformation. Unfortunately, at present, very little is known about these mechanisms and therefore it is instructive to find out more about the internal nature of granular samples by conducting suitable tests. In contrast to physical testing the method of investigation used in this study employs the Distinct Element Method. This is a computer based, iterative, time-dependent technique that allows the deformation of granular assemblies to be numerically simulated. By making assumptions regarding contact stiffnesses each individual contact force can be measured and by resolution particle centroid forces can be calculated. Then by dividing particle forces by their respective mass, particle centroid velocities and displacements are obtained by numerical integration. The Distinct Element Method is incorporated into a computer program 'Ball'. This program is effectively a numerical apparatus which forms a logical housing for this method and allows data input and output, and also provides testing control. By using this numerical apparatus tests have been carried out on disc assemblies and many new interesting observations regarding the micromechanical behaviour are revealed. In order to relate the observed microscopic mechanisms of deformation to the flow of the granular system two separate approaches have been used. Firstly a constitutive model has been developed which describes the yield function, flow rule and translation rule for regular assemblies of spheres and discs when subjected to coaxial deformation. Secondly statistical analyses have been carried out using data which was extracted from the simulation tests. These analyses define and quantify granular structure and then show how the force and velocity distributions use the structure to produce the corresponding stress and strain-rate tensors.

Typical case behaviour of spin systems in random graph and composite ensembles

This thesis includes analysis of disordered spin ensembles corresponding to Exact Cover, a multi-access channel problem, and composite models combining sparse and dense interactions. The satisfiability problem in Exact Cover is addressed using a statistical analysis of a simple branch and bound algorithm. The algorithm can be formulated in the large system limit as a branching process, for which critical properties can be analysed. Far from the critical point a set of differential equations may be used to model the process, and these are solved by numerical integration and exact bounding methods. The multi-access channel problem is formulated as an equilibrium statistical physics problem for the case of bit transmission on a channel with power control and synchronisation. A sparse code division multiple access method is considered and the optimal detection properties are examined in typical case by use of the replica method, and compared to detection performance achieved by interactive decoding methods. These codes are found to have phenomena closely resembling the well-understood dense codes. The composite model is introduced as an abstraction of canonical sparse and dense disordered spin models. The model includes couplings due to both dense and sparse topologies simultaneously. The new type of codes are shown to outperform sparse and dense codes in some regimes both in optimal performance, and in performance achieved by iterative detection methods in finite systems.

The role of pinning and instability in a class of non-equilibrium growth models

We study the dynamics of a growing crystalline facet where the growth mechanism is controlled by the geometry of the local curvature. A continuum model, in (2+1) dimensions, is developed in analogy with the Kardar-Parisi-Zhang (KPZ) model is considered for the purpose. Following standard coarse graining procedures, it is shown that in the large time, long distance limit, the continuum model predicts a curvature independent KPZ phase, thereby suppressing all explicit effects of curvature and local pinning in the system, in the "perturbative" limit. A direct numerical integration of this growth equation, in 1+1 dimensions, supports this observation below a critical parametric range, above which generic instabilities, in the form of isolated pillared structures lead to deviations from standard scaling behaviour. Possibilities of controlling this instability by introducing statistically "irrelevant" (in the sense of renormalisation groups) higher ordered nonlinearities have also been discussed.

Quasi-Monte Carlo Methods for some Linear Algebra Problems. Convergence and Complexity

We present quasi-Monte Carlo analogs of Monte Carlo methods for some linear algebra problems: solving systems of linear equations, computing extreme eigenvalues, and matrix inversion. Reformulating the problems as solving integral equations with a special kernels and domains permits us to analyze the quasi-Monte Carlo methods with bounds from numerical integration. Standard Monte Carlo methods for integration provide a convergence rate of O(N^(−1/2)) using N samples. Quasi-Monte Carlo methods use quasirandom sequences with the resulting convergence rate for numerical integration as good as O((logN)^k)N^(−1)). We have shown theoretically and through numerical tests that the use of quasirandom sequences improves both the magnitude of the error and the convergence rate of the considered Monte Carlo methods. We also analyze the complexity of considered quasi-Monte Carlo algorithms and compare them to the complexity of the analogous Monte Carlo and deterministic algorithms.

Polyharmonic Hardy Spaces on the Klein-Dirac Quadric with Application to Polyharmonic Interpolation and Cubature Formulas

2010 Mathematics Subject Classification: Primary 65D30, 32A35, Secondary 41A55.

Particle Trajectory Tracing and Computational Electromagnetics for Accelerator and Related Technologies

The main objective of this R&D work is to simulate particle beam optics in CV-28 Cyclotron of Instituto de Engenharia Nuclear – IEN/CNEN, as a support for improvements or optimization of this particle accelerator. Besides 2D magnetostatic field computation results, the authors present an alternative method for charged particle trajectories computation in electrostatic or magnetostatic fields. This task is approached by analytical computation of trajectories, by parts, considering constant fields within each finite element. This method has some advantages over numerical integration ones: numerical miscomputation of trajectories is avoided; stability problem is also avoided, for the magnetostatic field case. Some examples are presented, including positive ion extraction from cyclotrons with strip-foil. This latter technique is an interesting alternative for upgrading positive ion cyclotrons, such as CV-28 Cyclotron. The particle trajectory computation method presented in this work is of interest not only for cyclotrons, but for accelerator and related technology, in general.

Particle Trajectory Computation for Accelerators and Related Technologies

This work reports an alternative method for single non-relativistic charged particle trajectory computation in 2D electrostatic or magnetostatic fields. This task is approached by analytical computation of particle trajectory, by parts, considering the constant fields within each finite element. This method has some advantages over numerical integration ones: numerical miscomputation of trajectories, and stability problems can be avoided. Among the examples presented in this paper, an interesting alternative approach for positive ion extraction from cyclotrons is shown, using strip-foils. Other particle optics devices can benefit of a method such the one proposed in this paper, as beam bending devices, spectrometers, among others. This method can be extended for particle trajectory computation in 3D domains.