Computational Fluid Dynamics Expert System using Artificial Neural Networks

The design of a modern aircraft is based on three pillars: theoretical results, experimental test and computational simulations. As a results of this, Computational Fluid Dynamic (CFD) solvers are widely used in the aeronautical field. These solvers require the correct selection of many parameters in order to obtain successful results. Besides, the computational time spent in the simulation depends on the proper choice of these parameters. In this paper we create an expert system capable of making an accurate prediction of the number of iterations and time required for the convergence of a computational fluid dynamic (CFD) solver. Artificial neural network (ANN) has been used to design the expert system. It is shown that the developed expert system is capable of making an accurate prediction the number of iterations and time required for the convergence of a CFD solver.

Reduced order adaptive models for systems of PDEs using POD

In this work we propose a method to accelerate time dependent numerical solvers of systems of PDEs that require a high cost in computational time and memory. The method is based on the combined use of such numerical solver with a proper orthogonal decomposition, from which we identify modes, a Galerkin projection (that provides a reduced system of equations) and the integration of the reduced system, studying the evolution of the modal amplitudes. We integrate the reduced model until our a priori error estimator indicates that our approximation in not accurate. At this point we use again our original numerical code in a short time interval to adapt the POD manifold and continue then with the integration of the reduced model. Application will be made to two model problems: the Ginzburg-Landau equation in transient chaos conditions and the two-dimensional pulsating cavity problem, which describes the motion of liquid in a box whose upper wall is moving back and forth in a quasi-periodic fashion. Finally, we will discuss a way of improving the performance of the method using experimental data or information from numerical simulations

Numerical Error Prediction and its applications in CFD using tau-estimation

Nowadays, Computational Fluid Dynamics (CFD) solvers are widely used within the industry to model fluid flow phenomenons. Several fluid flow model equations have been employed in the last decades to simulate and predict forces acting, for example, on different aircraft configurations. Computational time and accuracy are strongly dependent on the fluid flow model equation and the spatial dimension of the problem considered. While simple models based on perfect flows, like panel methods or potential flow models can be very fast to solve, they usually suffer from a poor accuracy in order to simulate real flows (transonic, viscous). On the other hand, more complex models such as the full Navier- Stokes equations provide high fidelity predictions but at a much higher computational cost. Thus, a good compromise between accuracy and computational time has to be fixed for engineering applications. A discretisation technique widely used within the industry is the so-called Finite Volume approach on unstructured meshes. This technique spatially discretises the flow motion equations onto a set of elements which form a mesh, a discrete representation of the continuous domain. Using this approach, for a given flow model equation, the accuracy and computational time mainly depend on the distribution of nodes forming the mesh. Therefore, a good compromise between accuracy and computational time might be obtained by carefully defining the mesh. However, defining an optimal mesh for complex flows and geometries requires a very high level expertize in fluid mechanics and numerical analysis, and in most cases a simple guess of regions of the computational domain which might affect the most the accuracy is impossible. Thus, it is desirable to have an automatized remeshing tool, which is more flexible with unstructured meshes than its structured counterpart. However, adaptive methods currently in use still have an opened question: how to efficiently drive the adaptation ? Pioneering sensors based on flow features generally suffer from a lack of reliability, so in the last decade more effort has been made in developing numerical error-based sensors, like for instance the adjoint-based adaptation sensors. While very efficient at adapting meshes for a given functional output, the latter method is very expensive as it requires to solve a dual set of equations and computes the sensor on an embedded mesh. Therefore, it would be desirable to develop a more affordable numerical error estimation method. The current work aims at estimating the truncation error, which arises when discretising a partial differential equation. These are the higher order terms neglected in the construction of the numerical scheme. The truncation error provides very useful information as it is strongly related to the flow model equation and its discretisation. On one hand, it is a very reliable measure of the quality of the mesh, therefore very useful in order to drive a mesh adaptation procedure. On the other hand, it is strongly linked to the flow model equation, so that a careful estimation actually gives information on how well a given equation is solved, which may be useful in the context of _ -extrapolation or zonal modelling. The following work is organized as follows: Chap. 1 contains a short review of mesh adaptation techniques as well as numerical error prediction. In the first section, Sec. 1.1, the basic refinement strategies are reviewed and the main contribution to structured and unstructured mesh adaptation are presented. Sec. 1.2 introduces the definitions of errors encountered when solving Computational Fluid Dynamics problems and reviews the most common approaches to predict them. Chap. 2 is devoted to the mathematical formulation of truncation error estimation in the context of finite volume methodology, as well as a complete verification procedure. Several features are studied, such as the influence of grid non-uniformities, non-linearity, boundary conditions and non-converged numerical solutions. This verification part has been submitted and accepted for publication in the Journal of Computational Physics. Chap. 3 presents a mesh adaptation algorithm based on truncation error estimates and compares the results to a feature-based and an adjoint-based sensor (in collaboration with Jorge Ponsín, INTA). Two- and three-dimensional cases relevant for validation in the aeronautical industry are considered. This part has been submitted and accepted in the AIAA Journal. An extension to Reynolds Averaged Navier- Stokes equations is also included, where _ -estimation-based mesh adaptation and _ -extrapolation are applied to viscous wing profiles. The latter has been submitted in the Proceedings of the Institution of Mechanical Engineers, Part G: Journal of Aerospace Engineering. Keywords: mesh adaptation, numerical error prediction, finite volume Hoy en día, la Dinámica de Fluidos Computacional (CFD) es ampliamente utilizada dentro de la industria para obtener información sobre fenómenos fluidos. La Dinámica de Fluidos Computacional considera distintas modelizaciones de las ecuaciones fluidas (Potencial, Euler, Navier-Stokes, etc) para simular y predecir las fuerzas que actúan, por ejemplo, sobre una configuración de aeronave. El tiempo de cálculo y la precisión en la solución depende en gran medida de los modelos utilizados, así como de la dimensión espacial del problema considerado. Mientras que modelos simples basados en flujos perfectos, como modelos de flujos potenciales, se pueden resolver rápidamente, por lo general aducen de una baja precisión a la hora de simular flujos reales (viscosos, transónicos, etc). Por otro lado, modelos más complejos tales como el conjunto de ecuaciones de Navier-Stokes proporcionan predicciones de alta fidelidad, a expensas de un coste computacional mucho más elevado. Por lo tanto, en términos de aplicaciones de ingeniería se debe fijar un buen compromiso entre precisión y tiempo de cálculo. Una técnica de discretización ampliamente utilizada en la industria es el método de los Volúmenes Finitos en mallas no estructuradas. Esta técnica discretiza espacialmente las ecuaciones del movimiento del flujo sobre un conjunto de elementos que forman una malla, una representación discreta del dominio continuo. Utilizando este enfoque, para una ecuación de flujo dado, la precisión y el tiempo computacional dependen principalmente de la distribución de los nodos que forman la malla. Por consiguiente, un buen compromiso entre precisión y tiempo de cálculo se podría obtener definiendo cuidadosamente la malla, concentrando sus elementos en aquellas zonas donde sea estrictamente necesario. Sin embargo, la definición de una malla óptima para corrientes y geometrías complejas requiere un nivel muy alto de experiencia en la mecánica de fluidos y el análisis numérico, así como un conocimiento previo de la solución. Aspecto que en la mayoría de los casos no está disponible. Por tanto, es deseable tener una herramienta que permita adaptar los elementos de malla de forma automática, acorde a la solución fluida (remallado). Esta herramienta es generalmente más flexible en mallas no estructuradas que con su homóloga estructurada. No obstante, los métodos de adaptación actualmente en uso todavía dejan una pregunta abierta: cómo conducir de manera eficiente la adaptación. Sensores pioneros basados en las características del flujo en general, adolecen de una falta de fiabilidad, por lo que en la última década se han realizado grandes esfuerzos en el desarrollo numérico de sensores basados en el error, como por ejemplo los sensores basados en el adjunto. A pesar de ser muy eficientes en la adaptación de mallas para un determinado funcional, este último método resulta muy costoso, pues requiere resolver un doble conjunto de ecuaciones: la solución y su adjunta. Por tanto, es deseable desarrollar un método numérico de estimación de error más asequible. El presente trabajo tiene como objetivo estimar el error local de truncación, que aparece cuando se discretiza una ecuación en derivadas parciales. Estos son los términos de orden superior olvidados en la construcción del esquema numérico. El error de truncación proporciona una información muy útil sobre la solución: es una medida muy fiable de la calidad de la malla, obteniendo información que permite llevar a cabo un procedimiento de adaptación de malla. Está fuertemente relacionado al modelo matemático fluido, de modo que una estimación precisa garantiza la idoneidad de dicho modelo en un campo fluido, lo que puede ser útil en el contexto de modelado zonal. Por último, permite mejorar la precisión de la solución resolviendo un nuevo sistema donde el error local actúa como término fuente (_ -extrapolación). El presenta trabajo se organiza de la siguiente manera: Cap. 1 contiene una breve reseña de las técnicas de adaptación de malla, así como de los métodos de predicción de los errores numéricos. En la primera sección, Sec. 1.1, se examinan las estrategias básicas de refinamiento y se presenta la principal contribución a la adaptación de malla estructurada y no estructurada. Sec 1.2 introduce las definiciones de los errores encontrados en la resolución de problemas de Dinámica Computacional de Fluidos y se examinan los enfoques más comunes para predecirlos. Cap. 2 está dedicado a la formulación matemática de la estimación del error de truncación en el contexto de la metodología de Volúmenes Finitos, así como a un procedimiento de verificación completo. Se estudian varias características que influyen en su estimación: la influencia de la falta de uniformidad de la malla, el efecto de las no linealidades del modelo matemático, diferentes condiciones de contorno y soluciones numéricas no convergidas. Esta parte de verificación ha sido presentada y aceptada para su publicación en el Journal of Computational Physics. Cap. 3 presenta un algoritmo de adaptación de malla basado en la estimación del error de truncación y compara los resultados con sensores de featured-based y adjointbased (en colaboración con Jorge Ponsín del INTA). Se consideran casos en dos y tres dimensiones, relevantes para la validación en la industria aeronáutica. Este trabajo ha sido presentado y aceptado en el AIAA Journal. También se incluye una extensión de estos métodos a las ecuaciones RANS (Reynolds Average Navier- Stokes), en donde adaptación de malla basada en _ y _ -extrapolación son aplicados a perfiles con viscosidad de alas. Este último trabajo se ha presentado en los Actas de la Institución de Ingenieros Mecánicos, Parte G: Journal of Aerospace Engineering. Palabras clave: adaptación de malla, predicción del error numérico, volúmenes finitos

Prediction of wake effects on wind farm power production using a RANS approach. Part II. Offshore: Case studies from the UPWIND project

The estimation of power losses due to wind turbine wakes is crucial to understanding overall wind farm economics. This is especially true for large offshore wind farms, as it represents the primary source of losses in available power, given the regular arrangement of rotors, their generally largerdiameter and the lower ambient turbulence level, all of which conspire to dramatically affect wake expansion and, consequently, the power deficit. Simulation of wake effects in offshore wind farms (in reasonable computational time) is currently feasible using CFD tools. An elliptic CFD model basedon the actuator disk method and various RANS turbulence closure schemes is tested and validated using power ratios extracted from Horns Rev and Nysted wind farms, collected as part of the EU-funded UPWIND project. The primary focus of the present work is on turbulence modeling, as turbulent mixing is the main mechanism for flow recovery inside wind farms. A higher-order approach, based on the anisotropic RSM model, is tested to better take into account the imbalance in the length scales inside and outside of the wake, not well reproduced by current two-equation closure schemes.

Development of a wake model for wind farms based on an open source CFD solver. Strategies on parabolization and turbulence modeling

Wake effect represents one of the most important aspects to be analyzed at the engineering phase of every wind farm since it supposes an important power deficit and an increase of turbulence levels with the consequent decrease of the lifetime. It depends on the wind farm design, wind turbine type and the atmospheric conditions prevailing at the site. Traditionally industry has used analytical models, quick and robust, which allow carry out at the preliminary stages wind farm engineering in a flexible way. However, new models based on Computational Fluid Dynamics (CFD) are needed. These models must increase the accuracy of the output variables avoiding at the same time an increase in the computational time. Among them, the elliptic models based on the actuator disk technique have reached an extended use during the last years. These models present three important problems in case of being used by default for the solution of large wind farms: the estimation of the reference wind speed upstream of each rotor disk, turbulence modeling and computational time. In order to minimize the consequence of these problems, this PhD Thesis proposes solutions implemented under the open source CFD solver OpenFOAM and adapted for each type of site: a correction on the reference wind speed for the general elliptic models, the semi-parabollic model for large offshore wind farms and the hybrid model for wind farms in complex terrain. All the models are validated in terms of power ratios by means of experimental data derived from real operating wind farms.

Numerical methods for option pricing.

This thesis aims to introduce some fundamental concepts underlying option valuation theory including implementation of computational tools. In many cases analytical solution for option pricing does not exist, thus the following numerical methods are used: binomial trees, Monte Carlo simulations and finite difference methods. First, an algorithm based on Hull and Wilmott is written for every method. Then these algorithms are improved in different ways. For the binomial tree both speed and memory usage is significantly improved by using only one vector instead of a whole price storing matrix. Computational time in Monte Carlo simulations is reduced by implementing a parallel algorithm (in C) which is capable of improving speed by a factor which equals the number of processors used. Furthermore, MatLab code for Monte Carlo was made faster by vectorizing simulation process. Finally, obtained option values are compared to those obtained with popular finite difference methods, and it is discussed which of the algorithms is more appropriate for which purpose.

Analysis of the railway track as a spatially periodic structure

This article presents a new and computationally efficient method of analysis of a railway track modelled as a continuous beam of 2N spans supported by elastic vertical springs. The main feature of this method is its important reduction in computational effort with respect to standard matrix methods of structural analysis. In this article, the whole structure is considered to be a repetition of a single one. The analysis presented is applied to a simple railway track model, i.e. to a repetitive beam supported on vertical springs (sleepers). The proposed method of analysis is based on the general theory of spatially periodic structures. The main feature of this theory is the possibility to apply Discrete Fourier Transform (DFT) in order to reduce a large system of q(2N + 1) linear stiffness equilibrium equations to a set of 2N + 1 uncoupled systems of q equations each. In this way, a dramatic reduction of the computational effort of solving the large system of equations is achieved. This fact is particularly important in the analysis of railway track structures, in which N is a very large number (around several thousands), and q = 2, the vertical displacement and rotation, is very small. The proposed method allows us to easily obtain the exact solution given by Samartín [1], i.e. the continuous beam railway track response. The comparison between the proposed method and other methods of analysis of railway tracks, such as Lorente de Nó and Zimmermann-Timoshenko, clearly shows the accuracy of the obtained results for the proposed method, even for low values of N. In addition, identical results between the proposed and the Lorente methods have been found, although the proposed method seems to be of simpler application and computationally more efficient than the Lorente one. Small but significative differences occur between these two methods and the one developed by Zimmermann-Timoshenko. This article also presents a detailed sensitivity analysis of the vertical displacement of the sleepers. Although standard matrix methods of structural analysis can handle this railway model, one of the objectives of this article is to show the efficiency of DFT method with respect to standard matrix structural analysis. A comparative analysis between standard matrix structural analysis and the proposed method (DFT), in terms of computational time, input, output and also software programming, will be carried out. Finally, a URL link to a MatLab computer program list, based on the proposed method, is given

Interpretación geométrica de redes neuronales recurrentes discretas mediante grafos completos

El objetivo del presente trabajo de investigación es explorar nuevas técnicas de implementación, basadas en grafos, para las Redes de Neuronas, con el fin de simplificar y optimizar las arquitecturas y la complejidad computacional de las mismas. Hemos centrado nuestra atención en una clase de Red de Neuronas: las Redes de Neuronas Recursivas (RNR), también conocidas como redes de Hopfield. El problema de obtener la matriz sináptica asociada con una RNR imponiendo un determinado número de vectores como puntos fijos, no está en absoluto resuelto, el número de vectores prototipo que pueden ser almacenados en la red, cuando se utiliza la ley de Hebb, es bastante limitado, la red se satura rápidamente cuando se pretende almacenar nuevos prototipos. La ley de Hebb necesita, por tanto, ser revisada. Algunas aproximaciones dirigidas a solventar dicho problema, han sido ya desarrolladas. Nosotros hemos desarrollado una nueva aproximación en la forma de implementar una RNR en orden a solucionar estos problemas. La matriz sináptica es obtenida mediante la superposición de las componentes de los vectores prototipo, sobre los vértices de un Grafo, lo cual puede ser también interpretado como una coloración de dicho grafo. Cuando el periodo de entrenamiento se termina, la matriz de adyacencia del Grafo Resultante o matriz de pesos, presenta ciertas propiedades por las cuales dichas matrices serán llamadas tetraédricas. La energía asociada a cualquier estado de la red es representado por un punto (a,b) de R2. Cada uno de los puntos de energía asociados a estados que disten lo mismo del vector cero está localizado sobre la misma línea de energía de R2. El espacio de vectores de estado puede, por tanto, clasificarse en n clases correspondientes a cada una de las n diferentes distancias que puede tener cualquier vector al vector cero. La matriz (n x n) de pesos puede reducirse a un n-vector; de esta forma, tanto el tiempo de computación como el espacio de memoria requerido par almacenar los pesos, son simplificados y optimizados. En la etapa de recuperación, es introducido un vector de parámetros R2, éste es utilizado para controlar la capacidad de la red: probaremos que lo mayor es la componente a¡, lo menor es el número de puntos fijos pertenecientes a la línea de energía R¡. Una vez que la capacidad de la red ha sido controlada mediante este parámetro, introducimos otro parámetro, definido como la desviación del vector de pesos relativos, este parámetro sirve para disminuir ostensiblemente el número de parásitos. A lo largo de todo el trabajo, hemos ido desarrollando un ejemplo, el cual nos ha servido para ir corroborando los resultados teóricos, los algoritmos están escritos en un pseudocódigo, aunque a su vez han sido implamentados utilizando el paquete Mathematica 2.2., mostrándolos en un volumen suplementario al texto.---ABSTRACT---The aim of the present research is intended to explore new specifícation techniques of Neural Networks based on Graphs to be used in the optimization and simplification of Network Architectures and Computational Complexhy. We have focused our attention in a, well known, class of Neural Networks: the Recursive Neural Networks, also known as Hopfield's Neural Networks. The general problem of constructing the synaptic matrix associated with a Recursive Neural Network imposing some vectors as fixed points is fer for completery solved, the number of prototype vectors (learning patterns) which can be stored by Hebb's law is rather limited and the memory will thus quickly reach saturation if new prototypes are continuously acquired in the course of time. Hebb's law needs thus to be revised in order to allow new prototypes to be stored at the expense of the older ones. Some approaches related with this problem has been developed. We have developed a new approach of implementing a Recursive Neural Network in order to sob/e these kind of problems, the synaptic matrix is obtained superposing the components of the prototype vectors over the vértices of a Graph which may be interpreted as a coloring of the Graph. When training is finished the adjacency matrix of the Resulting Graph or matrix of weights presents certain properties for which it may be called a tetrahedral matrix The energy associated to any possible state of the net is represented as a point (a,b) in R2. Every one of the energy points associated with state-vectors having the same Hamming distance to the zero vector are located over the same energy Une in R2. The state-vector space may be then classified in n classes according to the n different possible distances firom any of the state-vectors to the zero vector The (n x n) matrix of weights may also be reduced to a n-vector of weights, in this way the computational time and the memory space required for obtaining the weights is optimized and simplified. In the recall stage, a parameter vectora is introduced, this parameter is used for controlling the capacity of the net: it may be proved that the bigger is the r, component of J, the lower is the number of fixed points located in the r¡ energy line. Once the capacity of the net has been controlled by the ex parameter, we introduced other parameter, obtained as the relative weight vector deviation parameter, in order to reduce the number of spurious states. All along the present text, we have also developed an example, which serves as a prove for the theoretical results, the algorithms are shown in a pseudocode language in the text, these algorithm so as the graphics have been developed also using the Mathematica 2.2. mathematical package which are shown in a supplementary volume of the text.

Fast 2D to 3D conversion using a clustering-based hierarchical search in a machine learning framework

Automatic 2D-to-3D conversion is an important application for filling the gap between the increasing number of 3D displays and the still scant 3D content. However, existing approaches have an excessive computational cost that complicates its practical application. In this paper, a fast automatic 2D-to-3D conversion technique is proposed, which uses a machine learning framework to infer the 3D structure of a query color image from a training database with color and depth images. Assuming that photometrically similar images have analogous 3D structures, a depth map is estimated by searching the most similar color images in the database, and fusing the corresponding depth maps. Large databases are desirable to achieve better results, but the computational cost also increases. A clustering-based hierarchical search using compact SURF descriptors to characterize images is proposed to drastically reduce search times. A significant computational time improvement has been obtained regarding other state-of-the-art approaches, maintaining the quality results.

Pulsatile blood flow interaction with arterial walls of aorta : autoregulation and impedance pressure boundary condition and its biomedical applications

Para las decisiones urgentes sobre intervenciones quirúrgicas en el sistema cardiovascular se necesitan simulaciones computacionales con resultados fiables y que consuman un tiempo de cálculo razonable. Durante años los investigadores han trabajado en diversos métodos numéricos de cálculo que resulten atractivos para los cirujanos. Estos métodos, precisos pero costosos desde el punto de vista del coste computacional, crean un desajuste entre la oferta de los ingenieros que realizan las simulaciones y los médicos que operan en el quirófano. Por otra parte, los métodos de cálculo más simplificados reducen el tiempo de cálculo pero pueden proporcionar resultados no realistas. El objetivo de esta tesis es combinar los conceptos de autorregulación e impedancia del sistema circulatorio, la interacción flujo sanguíneo-pared arterial y modelos geométricos idealizados tridimensionales de las arterias pero sin pérdida de realismo, con objeto de proponer una metodología de simulación que proporcione resultados correctos y completos, con tiempos de cálculo moderados. En las simulaciones numéricas, las condiciones de contorno basadas en historias de presión presentan inconvenientes por ser difícil conocerlas con detalle, y porque los resultados son muy sensibles ante pequeñas variaciones de dichas historias. La metodología propuesta se basa en los conceptos de autorregulación, para imponer la demanda de flujo aguas abajo del modelo en el ciclo cardiaco, y la impedancia, para representar el efecto que ejerce el flujo en el resto del sistema circulatorio sobre las arterias modeladas. De este modo las historias de presión en el contorno son resultados del cálculo, que se obtienen de manera iterativa. El método propuesto se aplica en una geometría idealizada del arco aórtico sin patologías y en otra geometría correspondiente a una disección Stanford de tipo A, considerando la interacción del flujo pulsátil con las paredes arteriales. El efecto de los tejidos circundantes también se incorpora en los modelos. También se hacen aplicaciones considerando la interacción en una geometría especifica de un paciente anciano que proviene de una tomografía computarizada. Finalmente se analiza una disección Stanford tipo B con tres modelos que incluyen la fenestración del saco. Clinicians demand fast and reliable numerical results of cardiovascular biomechanic simulations for their urgent pre-surgery decissions. Researchers during many years have work on different numerical methods in order to attract the clinicians' confidence to their colorful contours. Though precise but expensive and time-consuming methodologies create a gap between numerical biomechanics and hospital personnel. On the other hand, simulation simplifications with the aim of reduction in computational time may cause in production of unrealistic outcomes. The main objective of the current investigation is to combine ideas such as autoregulation, impedance, fluid-solid interaction and idealized geometries in order to propose a computationally cheap methodology without excessive or unrealistic simplifications. The pressure boundary conditions are critical and polemic in numerical simulations of cardiovascular system, in which a specific arterial site is of interest and the rest of the netwrok is neglected but represented by a boundary condition. The proposed methodology is a pressure boundary condition which takes advantage of numerical simplicity of application of an imposed pressure boundary condition on outlets, while it includes more sophisticated concepts such as autoregulation and impedance to gain more realistic results. Incorporation of autoregulation and impedance converts the pressure boundary conditions to an active and dynamic boundary conditions, receiving feedback from the results during the numerical calculations and comparing them with the physiological requirements. On the other hand, the impedance boundary condition defines the shapes of the pressure history curves applied at outlets. The applications of the proposed method are seen on idealized geometry of the healthy arotic arch as well as idealized Stanford type A dissection, considering the interaction of the arterial walls with the pulsatile blood flow. The effect of surrounding tissues is incorporated and studied in the models. The simulations continue with FSI analysis of a patient-specific CT scanned geometry of an old individual. Finally, inspiring of the statistic results of mortality rates in Stanford type B dissection, three models of fenestrated dissection sac is studied and discussed. Applying the developed boundary condition, an alternative hypothesis is proposed by the author with respect to the decrease in mortality rates in patients with fenestrations.

Método computacional para la aceleración de análisis paramétricos de modificaciones locales en estructuras complejas sometidas a cargas dinámicas

Dentro del análisis y diseño estructural surgen frecuentemente problemas de ingeniería donde se requiere el análisis dinámico de grandes modelos de elementos finitos que llegan a millones de grados de libertad y emplean volúmenes de datos de gran tamaño. La complejidad y dimensión de los análisis se dispara cuando se requiere realizar análisis paramétricos. Este problema se ha abordado tradicionalmente desde diversas perspectivas: en primer lugar, aumentando la capacidad tanto de cálculo como de memoria de los sistemas informáticos empleados en los análisis. En segundo lugar, se pueden simplificar los análisis paramétricos reduciendo su número o detalle y por último se puede recurrir a métodos complementarios a los elementos .nitos para la reducción de sus variables y la simplificación de su ejecución manteniendo los resultados obtenidos próximos al comportamiento real de la estructura. Se propone el empleo de un método de reducción que encaja en la tercera de las opciones y consiste en un análisis simplificado que proporciona una solución para la respuesta dinámica de una estructura en el subespacio modal complejo empleando un volumen de datos muy reducido. De este modo se pueden realizar análisis paramétricos variando múltiples parámetros, para obtener una solución muy aproximada al objetivo buscado. Se propone no solo la variación de propiedades locales de masa, rigidez y amortiguamiento sino la adición de grados de libertad a la estructura original para el cálculo de la respuesta tanto permanente como transitoria. Adicionalmente, su facilidad de implementación permite un control exhaustivo sobre las variables del problema y la implementación de mejoras como diferentes formas de obtención de los autovalores o la eliminación de las limitaciones de amortiguamiento en la estructura original. El objetivo del método se puede considerar similar a los que se obtienen al aplicar el método de Guyan u otras técnicas de reducción de modelos empleados en dinámica estructural. Sin embargo, aunque el método permite ser empleado en conjunción con otros para obtener las ventajas de ambos, el presente procedimiento no realiza la condensación del sistema de ecuaciones, sino que emplea la información del sistema de ecuaciones completa estudiando tan solo la respuesta en las variables apropiadas de los puntos de interés para el analista. Dicho interés puede surgir de la necesidad de obtener la respuesta de las grandes estructuras en unos puntos determinados o de la necesidad de modificar la estructura en zonas determinadas para cambiar su comportamiento (respuesta en aceleraciones, velocidades o desplazamientos) ante cargas dinámicas. Por lo tanto, el procedimiento está particularmente indicado para la selección del valor óptimo de varios parámetros en grandes estructuras (del orden de cientos de miles de modos) como pueden ser la localización de elementos introducidos, rigideces, masas o valores de amortiguamientos viscosos en estudios previos en los que diversas soluciones son planteadas y optimizadas, y que en el caso de grandes estructuras, pueden conllevar un número de simulaciones extremadamente elevado para alcanzar la solución óptima. Tras plantear las herramientas necesarias y desarrollar el procedimiento, se propone un caso de estudio para su aplicación al modelo de elementos .nitos del UAV MILANO desarrollado por el Instituto Nacional de Técnica Aeroespacial. A dicha estructura se le imponen ciertos requisitos al incorporar un equipo en aceleraciones en punta de ala izquierda y desplazamientos en punta de ala derecha en presencia de la sustentación producida por una ráfaga continua de viento de forma sinusoidal. La modificación propuesta consiste en la adición de un equipo en la punta de ala izquierda, bien mediante un anclaje rígido, bien unido mediante un sistema de reducción de la respuesta dinámica con propiedades de masa, rigidez y amortiguamiento variables. El estudio de los resultados obtenidos permite determinar la optimización de los parámetros del sistema de atenuación por medio de múltiples análisis dinámicos de forma que se cumplan de la mejor forma posible los requisitos impuestos con la modificación. Se comparan los resultados con los obtenidos mediante el uso de un programa comercial de análisis por el método de los elementos .nitos lográndose soluciones muy aproximadas entre el modelo completo y el reducido. La influencia de diversos factores como son el amortiguamiento modal de la estructura original, el número de modos retenidos en la truncatura o la precisión proporcionada por el barrido en frecuencia se analiza en detalle para, por último, señalar la eficiencia en términos de tiempo y volumen de datos de computación que ofrece el método propuesto en comparación con otras aproximaciones. Por lo tanto, puede concluirse que el método propuesto se considera una opción útil y eficiente para el análisis paramétrico de modificaciones locales en grandes estructuras. ABSTRACT When developing structural design and analysis some projects require dynamic analysis of large finite element models with millions of degrees of freedom which use large size data .les. The analysis complexity and size grow if a parametric analysis is required. This problem has been approached traditionally in several ways: one way is increasing the power and the storage capacity of computer systems involved in the analysis. Other obvious way is reducing the total amount of analyses and their details. Finally, complementary methods to finite element analysis can also be employed in order to limit the number of variables and to reduce the execution time keeping the results as close as possible to the actual behaviour of the structure. Following this third option, we propose a model reduction method that is based in a simplified analysis that supplies a solution for the dynamic response of the structure in the complex modal space using few data. Thereby, parametric analysis can be done varying multiple parameters so as to obtain a solution which complies with the desired objetive. We propose not only mass, stiffness and damping variations, but also addition of degrees of freedom to the original structure in order to calculate the transient and steady-state response. Additionally, the simple implementation of the procedure allows an in-depth control of the problem variables. Furthermore, improvements such as different ways to obtain eigenvectors or to remove damping limitations of the original structure are also possible. The purpose of the procedure is similar to that of using the Guyan or similar model order reduction techniques. However, in our method we do not perform a true model order reduction in the traditional sense. Furthermore, additional gains, which we do not explore herein, can be obtained through the combination of this method with traditional model-order reduction procedures. In our procedure we use the information of the whole system of equations is used but only those nodes of interest to the analyst are processed. That interest comes from the need to obtain the response of the structure at specific locations or from the need to modify the structure at some suitable positions in order to change its behaviour (acceleration, velocity or displacement response) under dynamic loads. Therefore, the procedure is particularly suitable for parametric optimization in large structures with >100000 normal modes such as position of new elements, stiffness, mass and viscous dampings in previous studies where different solutions are devised and optimized, and in the case of large structures, can carry an extremely high number of simulations to get the optimum solution. After the introduction of the required tools and the development of the procedure, a study case is proposed with use the finite element model (FEM) of the MILANO UAV developed by Instituto Nacional de Técnica Aeroespacial. Due to an equipment addition, certain acceleration and displacement requirements on left wing tip and right wing tip, respectively, are imposed. The structure is under a continuous sinusoidal wind gust which produces lift. The proposed modification consists of the addition of an equipment in left wing tip clamped through a rigid attachment or through a dynamic response reduction system with variable properties of mass, stiffness and damping. The analysis of the obtained results allows us to determine the optimized parametric by means of multiple dynamic analyses in a way such that the imposed requirements have been accomplished in the best possible way. The results achieved are compared with results from a commercial finite element analysis software, showing a good correlation. Influence of several factors such as the modal damping of the original structure, the number of modes kept in the modal truncation or the precission given by the frequency sweep is analyzed. Finally, the efficiency of the proposed method is addressed in tems of computational time and data size compared with other approaches. From the analyses performed, we can conclude that the proposed method is a useful and efficient option to perform parametric analysis of possible local modifications in large structures.

Profiling for run-time checking of computational properties and performance debugging in logic programs

Although several profiling techniques for identifying performance bottlenecks in logic programs have been developed, they are generally not automatic and in most cases they do not provide enough information for identifying the root causes of such bottlenecks. This complicates using their results for guiding performance improvement. We present a profiling method and tool that provides such explanations. Our profiler associates cost centers to certain program elements and can measure different types of resource-related properties that affect performance, preserving the precedence of cost centers in the cali graph. It includes an automatic method for detecting procedures that are performance bottlenecks. The profiling tool has been integrated in a previously developed run-time checking framework to allow verification of certain properties when they cannot be verified statically. The approach allows checking global computational properties which require complex instrumentation tracking information about previous execution states, such as, e.g., that the execution time accumulated by a given procedure is not greater than a given bound. We have built a prototype implementation, integrated it in the Ciao/CiaoPP system and successfully applied it to performance improvement, automatic optimization (e.g., resource-aware specialization of programs), run-time checking, and debugging of global computational properties (e.g., resource usage) in Prolog programs.

An approach to profiling for run-time checking of computational properties and performance debugging in logic programs.

Although several profiling techniques for identifying performance bottlenecks in logic programs have been developed, they are generally not automatic and in most cases they do not provide enough information for identifying the root causes of such bottlenecks. This complicates using their results for guiding performance improvement. We present a profiling method and tool that provides such explanations. Our profiler associates cost centers to certain program elements and can measure different types of resource-related properties that affect performance, preserving the precedence of cost centers in the call graph. It includes an automatic method for detecting procedures that are performance bottlenecks. The profiling tool has been integrated in a previously developed run-time checking framework to allow verification of certain properties when they cannot be verified statically. The approach allows checking global computational properties which require complex instrumentation tracking information about previous execution states, such as, e.g., that the execution time accumulated by a given procedure is not greater than a given bound. We have built a prototype implementation, integrated it in the Ciao/CiaoPP system and successfully applied it to performance improvement, automatic optimization (e.g., resource-aware specialization of programs), run-time checking, and debugging of global computational properties (e.g., resource usage) in Prolog programs.

Communication Middleware for Distributed Hard Real-Time Systems in Java

Distributed real-time embedded systems are becoming increasingly important to society. More demands will be made on them and greater reliance will be placed on the delivery of their services. A relevant subset of them is high-integrity or hard real-time systems, where failure can cause loss of life, environmental harm, or significant financial loss. Additionally, the evolution of communication networks and paradigms as well as the necessity of demanding processing power and fault tolerance, motivated the interconnection between electronic devices; many of the communications have the possibility of transferring data at a high speed. The concept of distributed systems emerged as systems where different parts are executed on several nodes that interact with each other via a communication network. Java’s popularity, facilities and platform independence have made it an interesting language for the real-time and embedded community. This was the motivation for the development of RTSJ (Real-Time Specification for Java), which is a language extension intended to allow the development of real-time systems. The use of Java in the development of high-integrity systems requires strict development and testing techniques. However, RTJS includes a number of language features that are forbidden in such systems. In the context of the HIJA project, the HRTJ (Hard Real-Time Java) profile was developed to define a robust subset of the language that is amenable to static analysis for high-integrity system certification. Currently, a specification under the Java community process (JSR- 302) is being developed. Its purpose is to define those capabilities needed to create safety critical applications with Java technology called Safety Critical Java (SCJ). However, neither RTSJ nor its profiles provide facilities to develop distributed realtime applications. This is an important issue, as most of the current and future systems will be distributed. The Distributed RTSJ (DRTSJ) Expert Group was created under the Java community process (JSR-50) in order to define appropriate abstractions to overcome this problem. Currently there is no formal specification. The aim of this thesis is to develop a communication middleware that is suitable for the development of distributed hard real-time systems in Java, based on the integration between the RMI (Remote Method Invocation) model and the HRTJ profile. It has been designed and implemented keeping in mind the main requirements such as the predictability and reliability in the timing behavior and the resource usage. iThe design starts with the definition of a computational model which identifies among other things: the communication model, most appropriate underlying network protocols, the analysis model, and a subset of Java for hard real-time systems. In the design, the remote references are the basic means for building distributed applications which are associated with all non-functional parameters and resources needed to implement synchronous or asynchronous remote invocations with real-time attributes. The proposed middleware separates the resource allocation from the execution itself by defining two phases and a specific threading mechanism that guarantees a suitable timing behavior. It also includes mechanisms to monitor the functional and the timing behavior. It provides independence from network protocol defining a network interface and modules. The JRMP protocol was modified to include two phases, non-functional parameters, and message size optimizations. Although serialization is one of the fundamental operations to ensure proper data transmission, current implementations are not suitable for hard real-time systems and there are no alternatives. This thesis proposes a predictable serialization that introduces a new compiler to generate optimized code according to the computational model. The proposed solution has the advantage of allowing us to schedule the communications and to adjust the memory usage at compilation time. In order to validate the design and the implementation a demanding validation process was carried out with emphasis in the functional behavior, the memory usage, the processor usage (the end-to-end response time and the response time in each functional block) and the network usage (real consumption according to the calculated consumption). The results obtained in an industrial application developed by Thales Avionics (a Flight Management System) and in exhaustive tests show that the design and the prototype are reliable for industrial applications with strict timing requirements. Los sistemas empotrados y distribuidos de tiempo real son cada vez más importantes para la sociedad. Su demanda aumenta y cada vez más dependemos de los servicios que proporcionan. Los sistemas de alta integridad constituyen un subconjunto de gran importancia. Se caracterizan por que un fallo en su funcionamiento puede causar pérdida de vidas humanas, daños en el medio ambiente o cuantiosas pérdidas económicas. La necesidad de satisfacer requisitos temporales estrictos, hace más complejo su desarrollo. Mientras que los sistemas empotrados se sigan expandiendo en nuestra sociedad, es necesario garantizar un coste de desarrollo ajustado mediante el uso técnicas adecuadas en su diseño, mantenimiento y certificación. En concreto, se requiere una tecnología flexible e independiente del hardware. La evolución de las redes y paradigmas de comunicación, así como la necesidad de mayor potencia de cómputo y de tolerancia a fallos, ha motivado la interconexión de dispositivos electrónicos. Los mecanismos de comunicación permiten la transferencia de datos con alta velocidad de transmisión. En este contexto, el concepto de sistema distribuido ha emergido como sistemas donde sus componentes se ejecutan en varios nodos en paralelo y que interactúan entre ellos mediante redes de comunicaciones. Un concepto interesante son los sistemas de tiempo real neutrales respecto a la plataforma de ejecución. Se caracterizan por la falta de conocimiento de esta plataforma durante su diseño. Esta propiedad es relevante, por que conviene que se ejecuten en la mayor variedad de arquitecturas, tienen una vida media mayor de diez anos y el lugar ˜ donde se ejecutan puede variar. El lenguaje de programación Java es una buena base para el desarrollo de este tipo de sistemas. Por este motivo se ha creado RTSJ (Real-Time Specification for Java), que es una extensión del lenguaje para permitir el desarrollo de sistemas de tiempo real. Sin embargo, RTSJ no proporciona facilidades para el desarrollo de aplicaciones distribuidas de tiempo real. Es una limitación importante dado que la mayoría de los actuales y futuros sistemas serán distribuidos. El grupo DRTSJ (DistributedRTSJ) fue creado bajo el proceso de la comunidad de Java (JSR-50) con el fin de definir las abstracciones que aborden dicha limitación, pero en la actualidad aun no existe una especificacion formal. El objetivo de esta tesis es desarrollar un middleware de comunicaciones para el desarrollo de sistemas distribuidos de tiempo real en Java, basado en la integración entre el modelo de RMI (Remote Method Invocation) y el perfil HRTJ. Ha sido diseñado e implementado teniendo en cuenta los requisitos principales, como la predecibilidad y la confiabilidad del comportamiento temporal y el uso de recursos. El diseño parte de la definición de un modelo computacional el cual identifica entre otras cosas: el modelo de comunicaciones, los protocolos de red subyacentes más adecuados, el modelo de análisis, y un subconjunto de Java para sistemas de tiempo real crítico. En el diseño, las referencias remotas son el medio básico para construcción de aplicaciones distribuidas las cuales son asociadas a todos los parámetros no funcionales y los recursos necesarios para la ejecución de invocaciones remotas síncronas o asíncronas con atributos de tiempo real. El middleware propuesto separa la asignación de recursos de la propia ejecución definiendo dos fases y un mecanismo de hebras especifico que garantiza un comportamiento temporal adecuado. Además se ha incluido mecanismos para supervisar el comportamiento funcional y temporal. Se ha buscado independencia del protocolo de red definiendo una interfaz de red y módulos específicos. También se ha modificado el protocolo JRMP para incluir diferentes fases, parámetros no funcionales y optimizaciones de los tamaños de los mensajes. Aunque la serialización es una de las operaciones fundamentales para asegurar la adecuada transmisión de datos, las actuales implementaciones no son adecuadas para sistemas críticos y no hay alternativas. Este trabajo propone una serialización predecible que ha implicado el desarrollo de un nuevo compilador para la generación de código optimizado acorde al modelo computacional. La solución propuesta tiene la ventaja que en tiempo de compilación nos permite planificar las comunicaciones y ajustar el uso de memoria. Con el objetivo de validar el diseño e implementación se ha llevado a cabo un exigente proceso de validación con énfasis en: el comportamiento funcional, el uso de memoria, el uso del procesador (tiempo de respuesta de extremo a extremo y en cada uno de los bloques funcionales) y el uso de la red (consumo real conforme al estimado). Los buenos resultados obtenidos en una aplicación industrial desarrollada por Thales Avionics (un sistema de gestión de vuelo) y en las pruebas exhaustivas han demostrado que el diseño y el prototipo son fiables para aplicaciones industriales con estrictos requisitos temporales.

Distributed models in P-Systems architectures to reduce computation time

Membrane systems are computational equivalent to Turing machines. However, their distributed and massively parallel nature obtains polynomial solutions opposite to traditional non-polynomial ones. At this point, it is very important to develop dedicated hardware and software implementations exploiting those two membrane systems features. Dealing with distributed implementations of P systems, the bottleneck communication problem has arisen. When the number of membranes grows up, the network gets congested. The purpose of distributed architectures is to reach a compromise between the massively parallel character of the system and the needed evolution step time to transit from one configuration of the system to the next one, solving the bottleneck communication problem. The goal of this paper is twofold. Firstly, to survey in a systematic and uniform way the main results regarding the way membranes can be placed on processors in order to get a software/hardware simulation of P-Systems in a distributed environment. Secondly, we improve some results about the membrane dissolution problem, prove that it is connected, and discuss the possibility of simulating this property in the distributed model. All this yields an improvement in the system parallelism implementation since it gets an increment of the parallelism of the external communication among processors. Proposed ideas improve previous architectures to tackle the communication bottleneck problem, such as reduction of the total time of an evolution step, increase of the number of membranes that could run on a processor and reduction of the number of processors.