Column Generation Algorithms for Nonlinear Optimization II: Numerical Investigations

García et al. present a class of column generation (CG) algorithms for nonlinear programs. Its main motivation from a theoretical viewpoint is that under some circumstances, finite convergence can be achieved, in much the same way as for the classic simplicial decomposition method; the main practical motivation is that within the class there are certain nonlinear column generation problems that can accelerate the convergence of a solution approach which generates a sequence of feasible points. This algorithm can, for example, accelerate simplicial decomposition schemes by making the subproblems nonlinear. This paper complements the theoretical study on the asymptotic and finite convergence of these methods given in [1] with an experimental study focused on their computational efficiency. Three types of numerical experiments are conducted. The first group of test problems has been designed to study the parameters involved in these methods. The second group has been designed to investigate the role and the computation of the prolongation of the generated columns to the relative boundary. The last one has been designed to carry out a more complete investigation of the difference in computational efficiency between linear and nonlinear column generation approaches. In order to carry out this investigation, we consider two types of test problems: the first one is the nonlinear, capacitated single-commodity network flow problem of which several large-scale instances with varied degrees of nonlinearity and total capacity are constructed and investigated, and the second one is a combined traffic assignment model

Gravitational actions upon a tether in a non-uniform gravity field with arbitrary number of zonal harmonics

We develop general closed-form expressions for the mutual gravitational potential, resultant and torque acting upon a rigid tethered system moving in a non-uniform gravity field produced by an attracting body with revolution symmetry, such that an arbitrary number of zonal harmonics is considered. The final expressions are series expansion in two small parameters related to the reference radius of the primary and the length of the tether, respectively, each of which are scaled by the mutual distance between their centers of mass. A few numerical experiments are performed to study the convergence behavior of the final expressions, and conclude that for high precision applications it might be necessary to take into account additional perturbation terms, which come from the mutual Two-Body interaction.

Some aspects of lubrication in heavy regimes: thermal effects, stability and turbulence

In this work, a combination of numerical methods applied to thermohydrodynamic lubrication problems with cavitation is presented. It should be emphasized the difficulty of the nonlinear mathematical coupled model involving a free boundary problem, but also the simplicity of the algorithms employed to solve it. So, finite element discretizations for the hydrodynamic and thermal equations combined with upwind techniques for the convection terms and duality methods for nonlinear features are proposed. Additionally, a model describing the movement of the shaft is provided. Considering the shaft as a rigid body this model will consist of an ODE system relating acceleration of the center of gravity and external and pressure loads. The numerical experiments of mechanical stability try to clarify the position of the neutral stability curve. Finally, a rotating machine for ship propulsion involving both axial and radial bearings operating with nonconventional lubricants (seawater to avoid environmental pollution) is analyzed by using laminar and turbulent inertial flows.

A subgrid viscosity Lagrance-Galerkin method for convection-diffusion problems

We present and analyze a subgrid viscosity Lagrange-Galerk in method that combines the subgrid eddy viscosity method proposed in W. Layton, A connection between subgrid scale eddy viscosity and mixed methods. Appl. Math. Comp., 133: 14 7-157, 2002, and a conventional Lagrange-Galerkin method in the framework of P1⊕ cubic bubble finite elements. This results in an efficient and easy to implement stabilized method for convection dominated convection diffusion reaction problems. Numerical experiments support the numerical analysis results and show that the new method is more accurate than the conventional Lagrange-Galerkin one.

On the flexural vibration of cylinders under axial loads:Numerical and experimental study

The flexural vibration of a homogeneous isotropic linearly elastic cylinder of any aspect ratio is analysed in this paper. Natural frequencies of a cylinder under uniformly distributed axial loads acting on its bases are calculated numerically by the Ritz method with terms of power series in the coordinate directions as approximating functions. The effect of axial loads on the flexural vibration cannot be described by applying infinitesimal strain theory, therefore, geometrically nonlinear strain–displacement relations with second-order terms are considered here. The natural frequencies of free–free, clamped–clamped, and sliding–sliding cylinders subjected to axial loads are calculated using the proposed three-dimensional Ritz approach and are compared with those obtained with the finite element method and the Bernoulli–Euler theory. Different experiments with cylinders axially compressed by a hydraulic press are carried out and the experimental results for the lowest flexural frequency are compared with the numerical results. An approach based on the Ritz formulation is proposed for the flexural vibration of a cylinder between the platens of the press with constraints varying with the intensity of the compression. The results show that for low compressions the cylinder behaves similarly to a sliding–sliding cylinder, whereas for high compressions the cylinder vibrates as a clamped–clamped one.

Experimental and numerical investigation of effect of sawn timber dimensions in ultrasonic velocity measurements of Spanish softwoods

Ultrasonic sound velocity measurements with hand-held equipment remain due to their simplicity among the most used methods for non-destructive grading of sawn woods, yet a dedicated normalization effort with respect to strength classes for Spanish species is still required. As part of an ongoing project with the aim of definition of standard testing methods, the effect of the dimensions of commonly tested Scots pine (Pinus sylvestris L.) timbers and equipment testing frequency on ultrasonic velocity were investigated. A dedicated full-wave finite-difference time-domain software allowed simulation of pulse propagation through timbers of representative length and section combinations. Sound velocity measurements vL were performed along the grain with the indirect method at 22 kHz and 45 kHz for grids of measurement points at specific distances. For sample sections larger than the cross-sectional wavelength ?RT, the simulated sound velocity vL converges to vL = (CL/?)0.5. For smaller square sections the sound velocity drops down to vL = (EL/?)0.5, where CL, EL and ? are the stiffness, E-modul and density, respectively. The experiments confirm a linear regression between time of flight and measurement distance even at less than two wavelength menor que2?L distance, the fitted sound speed values increased by 15% between the two tested frequencies.

Determination and analysis of plasma radiative properties for numerical simulations of laboratory radiative blast waves launched in xenon clusters

Radiative shock waves play a pivotal role in the transport energy into the stellar medium. This fact has led to many efforts to scale the astrophysical phenomena to accessible laboratory conditions and their study has been highlighted as an area requiring further experimental investigations. Low density material with high atomic mass is suitable to achieve radiative regime, and, therefore, low density xenon plasmas are commonly used for the medium in which the radiative shocks propagate. The knowledge of the plasma radiative properties is crucial for the correct understanding and for the hydrodynamic simulations of radiative shocks. In this work, we perform an analysis of the radiative properties of xenon plasmas in a range of matter densities and electron temperatures typically found in laboratory experiments of radiative shocks launched in xenon plasmas. Furthermore, for a particular experiment, our analysis is applied to make a diagnostics of the electron temperatures of the radiative shocks since they could not be experimentally measured

Experimental and numerical study of cracking of concrete due to reinforcement corrosion

La corrosión del acero es una de las patologías más importantes que afectan a las estructuras de hormigón armado que están expuestas a ambientes marinos o al ataque de sales fundentes. Cuando se produce corrosión, se genera una capa de óxido alrededor de la superficie de las armaduras, que ocupa un volumen mayor que el acero inicial; como consecuencia, el óxido ejerce presiones internas en el hormigón circundante, que lleva a la fisuración y, ocasionalmente, al desprendimiento del recubrimiento de hormigón. Durante los últimos años, numerosos estudios han contribuido a ampliar el conocimiento sobre el proceso de fisuración; sin embargo, aún existen muchas incertidumbres respecto al comportamiento mecánico de la capa de óxido, que es fundamental para predecir la fisuración. Por ello, en esta tesis se ha desarrollado y aplicado una metodología, para mejorar el conocimiento respecto al comportamiento del sistema acero-óxido-hormigón, combinando experimentos y simulaciones numéricas. Se han realizado ensayos de corrosión acelerada en condiciones de laboratorio, utilizando la técnica de corriente impresa. Con el objetivo de obtener información cercana a la capa de acero, como muestras se seleccionaron prismas de hormigón con un tubo de acero liso como armadura, que se diseñaron para conseguir la formación de una única fisura principal en el recubrimiento. Durante los ensayos, las muestras se equiparon con instrumentos especialmente diseñados para medir la variación de diámetro y volumen interior de los tubos, y se midió la apertura de la fisura principal utilizando un extensómetro comercial, adaptado a la geometría de las muestras. Las condiciones de contorno se diseñaron cuidadosamente para que los campos de corriente y deformación fuesen planos durante los ensayos, resultando en corrosión uniforme a lo largo del tubo, para poder reproducir los ensayos en simulaciones numéricas. Se ensayaron series con varias densidades de corriente y varias profundidades de corrosión. De manera complementaria, el comportamiento en fractura del hormigón se caracterizó en ensayos independientes, y se midió la pérdida gravimétrica de los tubos siguiendo procedimientos estándar. En todos los ensayos, la fisura principal creció muy despacio durante las primeras micras de profundidad de corrosión, pero después de una cierta profundidad crítica, la fisura se desarrolló completamente, con un aumento rápido de su apertura; la densidad de corriente influye en la profundidad de corrosión crítica. Las variaciones de diámetro interior y de volumen interior de los tubos mostraron tendencias diferentes entre sí, lo que indica que la deformación del tubo no fue uniforme. Después de la corrosión acelerada, las muestras se cortaron en rebanadas, que se utilizaron en ensayos post-corrosión. El patrón de fisuración se estudió a lo largo del tubo, en rebanadas que se impregnaron en vacío con resina y fluoresceína para mejorar la visibilidad de las fisuras bajo luz ultravioleta, y se estudió la presencia de óxido dentro de las grietas. En todas las muestras, se formó una fisura principal en el recubrimiento, infiltrada con óxido, y varias fisuras secundarias finas alrededor del tubo; el número de fisuras varió con la profundidad de corrosión de las muestras. Para muestras con la misma corrosión, el número de fisuras y su posición fue diferente entre muestras y entre secciones de una misma muestra, debido a la heterogeneidad del hormigón. Finalmente, se investigó la adherencia entre el acero y el hormigón, utilizando un dispositivo diseñado para empujar el tubo en el hormigón. Las curvas de tensión frente a desplazamiento del tubo presentaron un pico marcado, seguido de un descenso constante; la profundidad de corrosión y la apertura de fisura de las muestras influyeron notablemente en la tensión residual del ensayo. Para simular la fisuración del hormigón causada por la corrosión de las armaduras, se programó un modelo numérico. Éste combina elementos finitos con fisura embebida adaptable que reproducen la fractura del hormigón conforme al modelo de fisura cohesiva estándar, y elementos de interfaz llamados elementos junta expansiva, que se programaron específicamente para reproducir la expansión volumétrica del óxido y que incorporan su comportamiento mecánico. En el elemento junta expansiva se implementó un fenómeno de despegue, concretamente de deslizamiento y separación, que resultó fundamental para obtener localización de fisuras adecuada, y que se consiguió con una fuerte reducción de la rigidez tangencial y la rigidez en tracción del óxido. Con este modelo, se realizaron simulaciones de los ensayos, utilizando modelos bidimensionales de las muestras con elementos finitos. Como datos para el comportamiento en fractura del hormigón, se utilizaron las propiedades determinadas en experimentos. Para el óxido, inicialmente se supuso un comportamiento fluido, con deslizamiento y separación casi perfectos. Después, se realizó un ajuste de los parámetros del elemento junta expansiva para reproducir los resultados experimentales. Se observó que variaciones en la rigidez normal del óxido apenas afectaban a los resultados, y que los demás parámetros apenas afectaban a la apertura de fisura; sin embargo, la deformación del tubo resultó ser muy sensible a variaciones en los parámetros del óxido, debido a la flexibilidad de la pared de los tubos, lo que resultó fundamental para determinar indirectamente los valores de los parámetros constitutivos del óxido. Finalmente, se realizaron simulaciones definitivas de los ensayos. El modelo reprodujo la profundidad de corrosión crítica y el comportamiento final de las curvas experimentales; se comprobó que la variación de diámetro interior de los tubos está fuertemente influenciada por su posición relativa respecto a la fisura principal, en concordancia con los resultados experimentales. De la comparación de los resultados experimentales y numéricos, se pudo extraer información sobre las propiedades del óxido que de otra manera no habría podido obtenerse. Corrosion of steel is one of the main pathologies affecting reinforced concrete structures exposed to marine environments or to molten salt. When corrosion occurs, an oxide layer develops around the reinforcement surface, which occupies a greater volume than the initial steel; thus, it induces internal pressure on the surrounding concrete that leads to cracking and, eventually, to full-spalling of the concrete cover. During the last years much effort has been devoted to understand the process of cracking; however, there is still a lack of knowledge regarding the mechanical behavior of the oxide layer, which is essential in the prediction of cracking. Thus, a methodology has been developed and applied in this thesis to gain further understanding of the behavior of the steel-oxide-concrete system, combining experiments and numerical simulations. Accelerated corrosion tests were carried out in laboratory conditions, using the impressed current technique. To get experimental information close to the oxide layer, concrete prisms with a smooth steel tube as reinforcement were selected as specimens, which were designed to get a single main crack across the cover. During the tests, the specimens were equipped with instruments that were specially designed to measure the variation of inner diameter and volume of the tubes, and the width of the main crack was recorded using a commercial extensometer that was adapted to the geometry of the specimens. The boundary conditions were carefully designed so that plane current and strain fields were expected during the tests, resulting in nearly uniform corrosion along the length of the tube, so that the tests could be reproduced in numerical simulations. Series of tests were carried out with various current densities and corrosion depths. Complementarily, the fracture behavior of concrete was characterized in independent tests, and the gravimetric loss of the steel tubes was determined by standard means. In all the tests, the main crack grew very slowly during the first microns of corrosion depth, but after a critical corrosion depth it fully developed and opened faster; the current density influenced the critical corrosion depth. The variation of inner diameter and inner volume of the tubes had different trends, which indicates that the deformation of the tube was not uniform. After accelerated corrosion, the specimens were cut into slices, which were used in post-corrosion tests. The pattern of cracking along the reinforcement was investigated in slices that were impregnated under vacuum with resin containing fluorescein to enhance the visibility of cracks under ultraviolet lightening and a study was carried out to assess the presence of oxide into the cracks. In all the specimens, a main crack developed through the concrete cover, which was infiltrated with oxide, and several thin secondary cracks around the reinforcement; the number of cracks diminished with the corrosion depth of the specimen. For specimens with the same corrosion, the number of cracks and their position varied from one specimen to another and between cross-sections of a given specimen, due to the heterogeneity of concrete. Finally, the bond between the steel and the concrete was investigated, using a device designed to push the tubes of steel in the concrete. The curves of stress versus displacement of the tube presented a marked peak, followed by a steady descent, with notably influence of the corrosion depth and the crack width on the residual stress. To simulate cracking of concrete due to corrosion of the reinforcement, a numerical model was implemented. It combines finite elements with an embedded adaptable crack that reproduces cracking of concrete according to the basic cohesive model, and interface elements so-called expansive joint elements, which were specially designed to reproduce the volumetric expansion of oxide and incorporate its mechanical behavior. In the expansive joint element, a debonding effect was implemented consisting of sliding and separation, which was proved to be essential to achieve proper localization of cracks, and was achieved by strongly reducing the shear and the tensile stiffnesses of the oxide. With that model, simulations of the accelerated corrosion tests were carried out on 2- dimensional finite element models of the specimens. For the fracture behavior of concrete, the properties experimentally determined were used as input. For the oxide, initially a fluidlike behavior was assumed with nearly perfect sliding and separation; then the parameters of the expansive joint element were modified to fit the experimental results. Changes in the bulk modulus of the oxide barely affected the results and changes in the remaining parameters had a moderate effect on the predicted crack width; however, the deformation of the tube was very sensitive to variations in the parameters of oxide, due to the flexibility of the tube wall, which was crucial for indirect determination of the constitutive parameters of oxide. Finally, definitive simulations of the tests were carried out. The model reproduced the critical corrosion depth and the final behavior of the experimental curves; it was assessed that the variation of inner diameter of the tubes is highly influenced by its relative position with respect to the main crack, in accordance with the experimental observations. From the comparison of the experimental and numerical results, some properties of the mechanical behavior of the oxide were disclosed that otherwise could not have been measured.

Experimental and numerical assessment of the aeroelastic stability of blade pair packages

The stabilizing effect of grouping rotor blades in pairs has been assessed both, numerically and experimentally. The bending and torsion modes of a low aspect ratio high speed turbine cascade tested in the non-rotating test facility at EPFL (Ecole Polytechnique Fédérale de Lausanne) have been chosen as the case study. The controlled vibration of 20 blades in travelling wave form was performed by means of an electromagnetic excitation system, enabling the adjustement of the vibration amplitude and inter blade phase at a given frequency. Unsteady pressure transducers located along the blade mid-section were used to obtain the modulus and phase of the unsteady pressure caused by the airfoil motion. The stabilizing effect of the torsion mode was clearly observed both in the experiments and the simulations, however the effect of grouping the blades in pairs in the minimum damping at the tested frequency was marginal in the bending mode. A numerical tool was validated using the available experimental data and then used to extend the results at lower and more relevant reduced frequencies. It is shown that the stabilizing effect exists for the bending and torsion modes in the frequency range typical of low-pressure turbines. It is concluded that the stabilizing effect of this configuration is due to the shielding effect of the pressure side of the airfoil that defines the passage of the pair on the suction side of the same passage, since the relative motion between both is null. This effect is observed both in the experiments and simulations.

Experimental and numerical study of wind flow behind windbreaks

The shelter effect of a windbreak protects aggregate piles and provides a reduction of particle emissions in harbours. RANS (Reynolds-averaged Navier–Stokes equations) simulations using three variants of k–ε (standard k–ε, RNG k–ε and realizable k–ε) turbulence closure models have been performed to analyse wind flow characteristics behind an isolated fence located on a flat surface without roughness elements. The performance of the three turbulence models has been assessed by wind tunnel experiments. Cases of fences with different porosities (φ) have been evaluated using wind tunnel experiments as well as numerical simulations. The aim is to determine an optimum porosity for sheltering effect of an isolated windbreak. A value of 0.35 was found as the optimum value among the studied porosities (φ=0, 0.1, 0.24, 0.35, 0.4, 0.5).

Flow around a hemisphere-cylinder at high angle of attack and low Reynolds number. Part I: Experimental and numerical investigation

Three-dimensional Direct Numerical Simulations combined with Particle Image Velocimetry experiments have been performed on a hemisphere-cylinder at Reynolds number 1000 and angle of attack 20◦. At these flow conditions, a pair of vortices, so-called “horn” vortices, are found to be associated with flow separation. In order to understand the highly complex phenomena associated with this fully threedimensional massively separated flow, different structural analysis techniques have been employed: Proper Orthogonal and Dynamic Mode Decompositions, POD and DMD, respectively, as well as criticalpoint theory. A single dominant frequency associated with the von Karman vortex shedding has been identified in both the experimental and the numerical results. POD and DMD modes associated with this frequency were recovered in the analysis. Flow separation was also found to be intrinsically linked to the observed modes. On the other hand, critical-point theory has been applied in order to highlight possible links of the topology patterns over the surface of the body with the computed modes. Critical points and separation lines on the body surface show in detail the presence of different flow patterns in the base flow: a three-dimensional separation bubble and two pairs of unsteady vortices systems, the horn vortices, mentioned before, and the so-called “leeward” vortices. The horn vortices emerge perpendicularly from the body surface at the separation region. On the other hand, the leeward vortices are originated downstream of the separation bubble, as a result of the boundary layer separation. The frequencies associated with these vortical structures have been quantified.