Reducing Cache Hierarchy Energy Consumption by Predicting Forwarding and Disabling Associative Sets

The first level data cache un modern processors has become a major consumer of energy due to its increasing size and high frequency access rate. In order to reduce this high energy con sumption, we propose in this paper a straightforward filtering technique based on a highly accurate forwarding predictor. Specifically, a simple structure predicts whether a load instruction will obtain its corresponding data via forwarding from the load-store structure -thus avoiding the data cache access - or if it will be provided by the data cache. This mechanism manages to reduce the data cache energy consumption by an average of 21.5% with a negligible performance penalty of less than 0.1%. Furthermore, in this paper we focus on the cache static energy consumption too by disabling a portin of sets of the L2 associative cache. Overall, when merging both proposals, the combined L1 and L2 total energy consumption is reduced by an average of 29.2% with a performance penalty of just 0.25%. Keywords: Energy consumption; filtering; forwarding predictor; cache hierarchy

A multi-criteria optimization and decision-making approach for improvement of food engineering processes.

The objective of this study was to propose a multi-criteria optimization and decision-making technique to solve food engineering problems. This technique was demostrated using experimental data obtained on osmotic dehydratation of carrot cubes in a sodium chloride solution. The Aggregating Functions Approach, the Adaptive Random Search Algorithm, and the Penalty Functions Approach were used in this study to compute the initial set of non-dominated or Pareto-optimal solutions. Multiple non-linear regression analysis was performed on a set of experimental data in order to obtain particular multi-objective functions (responses), namely water loss, solute gain, rehydration ratio, three different colour criteria of rehydrated product, and sensory evaluation (organoleptic quality). Two multi-criteria decision-making approaches, the Analytic Hierarchy Process (AHP) and the Tabular Method (TM), were used simultaneously to choose the best alternative among the set of non-dominated solutions. The multi-criteria optimization and decision-making technique proposed in this study can facilitate the assessment of criteria weights, giving rise to a fairer, more consistent, and adequate final compromised solution or food process. This technique can be useful to food scientists in research and education, as well as to engineers involved in the improvement of a variety of food engineering processes.

3D Frangi-based lung vessel enhancement filter penalizing airways

This paper describes a fully automatic simultaneous lung vessel and airway enhancement filter. The approach consists of a Frangi-based multiscale vessel enhancement filtering specifically designed for lung vessel and airway detection, where arteries and veins have high contrast with respect to the lung parenchyma, and airway walls are hollow tubular structures with a non negative response using the classical Frangi's filter. The features extracted from the Hessian matrix are used to detect centerlines and approximate walls of airways, decreasing the filter response in those areas by applying a penalty function to the vesselness measure. We validate the segmentation method in 20 CT scans with different pathological states within the VESSEL12 challenge framework. Results indicate that our approach obtains good results, decreasing the number of false positives in airway walls.

Cross-Products LASSO

Negative co-occurrence is a common phenomenon in many signal processing applications. In some cases the signals involved are sparse, and this information can be exploited to recover them. In this paper, we present a sparse learning approach that explicitly takes into account negative co-occurrence. This is achieved by adding a novel penalty term to the LASSO cost function based on the cross-products between the reconstruction coefficients. Although the resulting optimization problem is non-convex, we develop a new and efficient method for solving it based on successive convex approximations. Results on synthetic data, for both complete and overcomplete dictionaries, are provided to validate the proposed approach.

Dynamic behaviour of underspanned suspension road bridges under traffic loads

Underspanned suspension bridges are structures with important economical and aesthetic advantages, due to their high structural efficiency. However, road bridges of this typology are still uncommon because of limited knowledge about this structural system. In particular, there remains some uncertainty over the dynamic behaviour of these bridges, due to their extreme lightness. The vibrations produced by vehicles crossing the viaduct are one of the main concerns. In this work, traffic-induced dynamic effects on this kind of viaduct are addressed by means of vehicle-bridge dynamic interaction models. A finite element method is used for the structure, and multibody dynamic models for the vehicles, while interaction is represented by means of the penalty method. Road roughness is included in this model in such a way that the fact that profiles under left and right tyres are different, but not independent, is taken into account. In addition, free software {PRPgenerator) to generate these profiles is presented in this paper. The structural dynamic sensitivity of underspanned suspension bridges was found to be considerable, as well as the dynamic amplification factors and deck accelerations. It was also found that vehicle speed has a relevant influence on the results. In addition, the impact of bridge deformation on vehicle vibration was addressed, and the effect on the comfort of vehicle users was shown to be negligible.

An optimal yet fast pruning algorithm to reduce latency in multiview prediction structures

We propose a new algorithm for the design of prediction structures with low delay and limited penalty in the rate-distortion performance for multiview video coding schemes. This algorithm constitutes one of the elements of a framework for the analysis and optimization of delay in multiview coding schemes that is based in graph theory. The objective of the algorithm is to find the best combination of prediction dependencies to prune from a multiview prediction structure, given a number of cuts. Taking into account the properties of the graph-based analysis of the encoding delay, the algorithm is able to find the best prediction dependencies to eliminate from an original prediction structure, while limiting the number of cut combinations to evaluate. We show that this algorithm obtains optimum results in the reduction of the encoding latency with a lower computational complexity than exhaustive search alternatives.

Optical signal impairment study of cascaded optical filters in 40 Gbps DQPSK and 100 Gbps PM-DQPSK systems

Optical filters are crucial elements in optical communications. The influence of cascaded filters in the optical signal will affect the communications quality seriously. In this paper we will study and simulate the optical signal impairment caused by different kinds of filters which include Butterworth, Bessel, Fiber Bragg Grating (FBG) and Fabry-Perot (FP). Optical signal impairment is analyzed from an Eye Opening Penalty (EOP) and optical spectrum point of view. The simulation results show that when the center frequency of all filters aligns with the laser’s frequency, the Butterworth has the smallest influence to the signal while the F-P has the biggest. With a -1dB EOP, the amount of cascaded Butterworth optical filters with a bandwidth of 50 GHz is 18 in 40 Gbps NRZ-DQPSK systems and 12 in 100 Gbps PMNRZ- DQPSK systems. The value is reduced to 9 and 6 respectively for Febry-Perot optical filters. In the situation of frequency misalignment, the impairment caused by filters is more serious. Our research shows that with a frequency deviation of 5 GHz, only 12 and 9 Butterworth optical filters can be cascaded in 40 Gbps NRZ-DQPSK and 100 Gbps PM-NRZ-DQPSK systems respectively. We also study the signal impairment caused by different orders of the Butterworth filter model. Our study shows that although the higher-order has a smaller clipping effect in the transmission spectrum, it will introduce a more serious phase ripple which seriously affects the signal. Simulation result shows that the 2nd order Butterworth filter has the best performance.

Simulation of an Integrated Gasification Combined Cycle with Chemical-Looping Combustion and CO2 sequestration

Chemical-looping combustion allows an integration of CO2 capture in a thermal power plant without energy penalty; secondly, a less exergy destruction in the combustion chemical transformation is achieved, leading to a greater overall thermal efficiency. This paper focus on the study of the energetic performance of this concept of combustion in an integrated gasification combined cycle power plant when synthesis gas is used as fuel for the gas turbines. After thermodynamic modelling and optimization of some cycle parameters, the power plant performance is evaluated under diverse working conditions and compared to a conventional integrated gasification combined cycle with precombustion capture. Energy savings in CO2 capture and storage has been quantified. The overall efficiency increase is found to be significant and even notable, reaching values of around 7%. In order to analyze the influence of syngas composition on the results, different H2-content fuels are considered.

Análisis energético de las centrales de ciclo combinado con gasificación integrada y combustión mediante lazo químico

La combustión mediante lazo químico permite una integración de captura de CO2 en una planta térmica de potencia sin penalización energética. Se alcanza una menor destrucción exergética, alcanzándose un rendimiento térmico idóneo. Este proyecto se centra en el estudio del rendimiento energético en una planta de potencia con gasificación integrada cuando la cámara de combustión de las turbinas de gas se sustituye por un sistema de combustión CLC. Después del diseño termodinámico y optimización de algunos parámetros del ciclo, el rendimiento de la planta de potencia es evaluado bajo distintas condiciones de trabajo y comparado con el que se obtendría en una planta de ciclo combinado convencional con gasificación integrada y captura pre-combustión. Abstract Chemical-looping combustion allows an integration of CO2 capture in a thermal power plant without energy penalty. A less exergy destruction in the combustion chemical transformation is achieved, leading to a greater overall thermal efficiency. This Project focus on the study of the energetic performance in a cycle power plant with integrated gasification when the chamber combustion is supplied by a combustion system CLC. After thermodynamic modeling and optimization of some cycle parameters, the power plant performance is evaluated under diverse working conditions and compared to a conventional integrated gasification combined cycle with pre-combustion capture.

Modelos de elementos finitos explícitos para explosiones en estructuras reticuladas de hormigón armado : aplicaciones al estudio del colapso de edificios

La frecuencia con la que se producen explosiones sobre edificios, ya sean accidentales o intencionadas, es reducida, pero sus efectos pueden ser catastróficos. Es deseable poder predecir de forma suficientemente precisa las consecuencias de estas acciones dinámicas sobre edificaciones civiles, entre las cuales las estructuras reticuladas de hormigón armado son una tipología habitual. En esta tesis doctoral se exploran distintas opciones prácticas para el modelado y cálculo numérico por ordenador de estructuras de hormigón armado sometidas a explosiones. Se emplean modelos numéricos de elementos finitos con integración explícita en el tiempo, que demuestran su capacidad efectiva para simular los fenómenos físicos y estructurales de dinámica rápida y altamente no lineales que suceden, pudiendo predecir los daños ocasionados tanto por la propia explosión como por el posible colapso progresivo de la estructura. El trabajo se ha llevado a cabo empleando el código comercial de elementos finitos LS-DYNA (Hallquist, 2006), desarrollando en el mismo distintos tipos de modelos de cálculo que se pueden clasificar en dos tipos principales: 1) modelos basados en elementos finitos de continuo, en los que se discretiza directamente el medio continuo mediante grados de libertad nodales de desplazamientos; 2) modelos basados en elementos finitos estructurales, mediante vigas y láminas, que incluyen hipótesis cinemáticas para elementos lineales o superficiales. Estos modelos se desarrollan y discuten a varios niveles distintos: 1) a nivel del comportamiento de los materiales, 2) a nivel de la respuesta de elementos estructurales tales como columnas, vigas o losas, y 3) a nivel de la respuesta de edificios completos o de partes significativas de los mismos. Se desarrollan modelos de elementos finitos de continuo 3D muy detallados que modelizan el hormigón en masa y el acero de armado de forma segregada. El hormigón se representa con un modelo constitutivo del hormigón CSCM (Murray et al., 2007), que tiene un comportamiento inelástico, con diferente respuesta a tracción y compresión, endurecimiento, daño por fisuración y compresión, y rotura. El acero se representa con un modelo constitutivo elastoplástico bilineal con rotura. Se modeliza la geometría precisa del hormigón mediante elementos finitos de continuo 3D y cada una de las barras de armado mediante elementos finitos tipo viga, con su posición exacta dentro de la masa de hormigón. La malla del modelo se construye mediante la superposición de los elementos de continuo de hormigón y los elementos tipo viga de las armaduras segregadas, que son obligadas a seguir la deformación del sólido en cada punto mediante un algoritmo de penalización, simulando así el comportamiento del hormigón armado. En este trabajo se denominarán a estos modelos simplificadamente como modelos de EF de continuo. Con estos modelos de EF de continuo se analiza la respuesta estructural de elementos constructivos (columnas, losas y pórticos) frente a acciones explosivas. Asimismo se han comparado con resultados experimentales, de ensayos sobre vigas y losas con distintas cargas de explosivo, verificándose una coincidencia aceptable y permitiendo una calibración de los parámetros de cálculo. Sin embargo estos modelos tan detallados no son recomendables para analizar edificios completos, ya que el elevado número de elementos finitos que serían necesarios eleva su coste computacional hasta hacerlos inviables para los recursos de cálculo actuales. Adicionalmente, se desarrollan modelos de elementos finitos estructurales (vigas y láminas) que, con un coste computacional reducido, son capaces de reproducir el comportamiento global de la estructura con una precisión similar. Se modelizan igualmente el hormigón en masa y el acero de armado de forma segregada. El hormigón se representa con el modelo constitutivo del hormigón EC2 (Hallquist et al., 2013), que también presenta un comportamiento inelástico, con diferente respuesta a tracción y compresión, endurecimiento, daño por fisuración y compresión, y rotura, y se usa en elementos finitos tipo lámina. El acero se representa de nuevo con un modelo constitutivo elastoplástico bilineal con rotura, usando elementos finitos tipo viga. Se modeliza una geometría equivalente del hormigón y del armado, y se tiene en cuenta la posición relativa del acero dentro de la masa de hormigón. Las mallas de ambos se unen mediante nodos comunes, produciendo una respuesta conjunta. En este trabajo se denominarán a estos modelos simplificadamente como modelos de EF estructurales. Con estos modelos de EF estructurales se simulan los mismos elementos constructivos que con los modelos de EF de continuo, y comparando sus respuestas estructurales frente a explosión se realiza la calibración de los primeros, de forma que se obtiene un comportamiento estructural similar con un coste computacional reducido. Se comprueba que estos mismos modelos, tanto los modelos de EF de continuo como los modelos de EF estructurales, son precisos también para el análisis del fenómeno de colapso progresivo en una estructura, y que se pueden utilizar para el estudio simultáneo de los daños de una explosión y el posterior colapso. Para ello se incluyen formulaciones que permiten considerar las fuerzas debidas al peso propio, sobrecargas y los contactos de unas partes de la estructura sobre otras. Se validan ambos modelos con un ensayo a escala real en el que un módulo con seis columnas y dos plantas colapsa al eliminar una de sus columnas. El coste computacional del modelo de EF de continuo para la simulación de este ensayo es mucho mayor que el del modelo de EF estructurales, lo cual hace inviable su aplicación en edificios completos, mientras que el modelo de EF estructurales presenta una respuesta global suficientemente precisa con un coste asumible. Por último se utilizan los modelos de EF estructurales para analizar explosiones sobre edificios de varias plantas, y se simulan dos escenarios con cargas explosivas para un edificio completo, con un coste computacional moderado. The frequency of explosions on buildings whether they are intended or accidental is small, but they can have catastrophic effects. Being able to predict in a accurate enough manner the consequences of these dynamic actions on civil buildings, among which frame-type reinforced concrete buildings are a frequent typology is desirable. In this doctoral thesis different practical options for the modeling and computer assisted numerical calculation of reinforced concrete structures submitted to explosions are explored. Numerical finite elements models with explicit time-based integration are employed, demonstrating their effective capacity in the simulation of the occurring fast dynamic and highly nonlinear physical and structural phenomena, allowing to predict the damage caused by the explosion itself as well as by the possible progressive collapse of the structure. The work has been carried out with the commercial finite elements code LS-DYNA (Hallquist, 2006), developing several types of calculation model classified in two main types: 1) Models based in continuum finite elements in which the continuous medium is discretized directly by means of nodal displacement degrees of freedom; 2) Models based on structural finite elements, with beams and shells, including kinematic hypothesis for linear and superficial elements. These models are developed and discussed at different levels: 1) material behaviour, 2) response of structural elements such as columns, beams and slabs, and 3) response of complete buildings or significative parts of them. Very detailed 3D continuum finite element models are developed, modeling mass concrete and reinforcement steel in a segregated manner. Concrete is represented with a constitutive concrete model CSCM (Murray et al., 2007), that has an inelastic behaviour, with different tension and compression response, hardening, cracking and compression damage and failure. The steel is represented with an elastic-plastic bilinear model with failure. The actual geometry of the concrete is modeled with 3D continuum finite elements and every and each of the reinforcing bars with beam-type finite elements, with their exact position in the concrete mass. The mesh of the model is generated by the superposition of the concrete continuum elements and the beam-type elements of the segregated reinforcement, which are made to follow the deformation of the solid in each point by means of a penalty algorithm, reproducing the behaviour of reinforced concrete. In this work these models will be called continuum FE models as a simplification. With these continuum FE models the response of construction elements (columns, slabs and frames) under explosive actions are analysed. They have also been compared with experimental results of tests on beams and slabs with various explosive charges, verifying an acceptable coincidence and allowing a calibration of the calculation parameters. These detailed models are however not advised for the analysis of complete buildings, as the high number of finite elements necessary raises its computational cost, making them unreliable for the current calculation resources. In addition to that, structural finite elements (beams and shells) models are developed, which, while having a reduced computational cost, are able to reproduce the global behaviour of the structure with a similar accuracy. Mass concrete and reinforcing steel are also modeled segregated. Concrete is represented with the concrete constitutive model EC2 (Hallquist et al., 2013), which also presents an inelastic behaviour, with a different tension and compression response, hardening, compression and cracking damage and failure, and is used in shell-type finite elements. Steel is represented once again with an elastic-plastic bilineal with failure constitutive model, using beam-type finite elements. An equivalent geometry of the concrete and the steel is modeled, considering the relative position of the steel inside the concrete mass. The meshes of both sets of elements are bound with common nodes, therefore producing a joint response. These models will be called structural FE models as a simplification. With these structural FE models the same construction elements as with the continuum FE models are simulated, and by comparing their response under explosive actions a calibration of the former is carried out, resulting in a similar response with a reduced computational cost. It is verified that both the continuum FE models and the structural FE models are also accurate for the analysis of the phenomenon of progressive collapse of a structure, and that they can be employed for the simultaneous study of an explosion damage and the resulting collapse. Both models are validated with an experimental full-scale test in which a six column, two floors module collapses after the removal of one of its columns. The computational cost of the continuum FE model for the simulation of this test is a lot higher than that of the structural FE model, making it non-viable for its application to full buildings, while the structural FE model presents a global response accurate enough with an admissible cost. Finally, structural FE models are used to analyze explosions on several story buildings, and two scenarios are simulated with explosive charges for a full building, with a moderate computational cost.

Optimización del consumo de combustible de un vehículo dependiendo de las condiciones de la carretera

El consumo de combustible en un automóvil es una característica que se intenta mejorar continuamente debido a los precios del carburante y a la creciente conciencia medioambiental. Esta tesis doctoral plantea un algoritmo de optimización del consumo que tiene en cuenta las especificaciones técnicas del vehículo, el perfil de orografía de la carretera y el tráfico presente en ella. El algoritmo de optimización calcula el perfil de velocidad óptima que debe seguir el vehículo para completar un recorrido empleando un tiempo de viaje especificado. El cálculo del perfil de velocidad óptima considera los valores de pendiente de la carretera así como también las condiciones de tráfico vehicular de la franja horaria en que se realiza el recorrido. El algoritmo de optimización reacciona ante condiciones de tráfico cambiantes y adapta continuamente el perfil óptimo de velocidad para que el vehículo llegue al destino cumpliendo el horario de llegada establecido. La optimización de consumo es aplicada en vehículos convencionales de motor de combustión interna y en vehículos híbridos tipo serie. Los datos de consumo utilizados por el algoritmo de optimización se obtienen mediante la simulación de modelos cuasi-estáticos de los vehículos. La técnica de minimización empleada por el algoritmo es la Programación Dinámica. El algoritmo divide la optimización del consumo en dos partes claramente diferenciadas y aplica la Programación Dinámica sobre cada una de ellas. La primera parte corresponde a la optimización del consumo del vehículo en función de las condiciones de tráfico. Esta optimización calcula un perfil de velocidad promedio que evita, cuando es posible, las retenciones de tráfico. El tiempo de viaje perdido durante una retención de tráfico debe recuperarse a través de un aumento posterior de la velocidad promedio que incrementaría el consumo del vehículo. La segunda parte de la optimización es la encargada del cálculo de la velocidad óptima en función de la orografía y del tiempo de viaje disponible. Dado que el consumo de combustible del vehículo se incrementa cuando disminuye el tiempo disponible para finalizar un recorrido, esta optimización utiliza factores de ponderación para modular la influencia que tiene cada una de estas dos variables en el proceso de minimización. Aunque los factores de ponderación y la orografía de la carretera condicionan el nivel de ahorro de la optimización, los perfiles de velocidad óptima calculados logran ahorros de consumo respecto de un perfil de velocidad constante que obtiene el mismo tiempo de recorrido. Las simulaciones indican que el ahorro de combustible del vehículo convencional puede lograr hasta un 8.9% mientras que el ahorro de energía eléctrica del vehículo híbrido serie un 2.8%. El algoritmo fusiona la optimización en función de las condiciones del tráfico y la optimización en función de la orografía durante el cálculo en tiempo real del perfil óptimo de velocidad. La optimización conjunta se logra cuando el perfil de velocidad promedio resultante de la optimización en función de las condiciones de tráfico define los valores de los factores de ponderación de la optimización en función de la orografía. Aunque el nivel de ahorro de la optimización conjunta depende de las condiciones de tráfico, de la orografía, del tiempo de recorrido y de las características propias del vehículo, las simulaciones indican ahorros de consumo superiores al 6% en ambas clases de vehículo respecto a optimizaciones que no logran evitar retenciones de tráfico en la carretera. ABSTRACT Fuel consumption of cars is a feature that is continuously being improved due to the fuel price and an increasing environmental awareness. This doctoral dissertation describes an optimization algorithm to decrease the fuel consumption taking into account the technical specifications of the vehicle, the terrain profile of the road and the traffic conditions of the trip. The algorithm calculates the optimal speed profile that completes a trip having a specified travel time. This calculation considers the road slope and the expected traffic conditions during the trip. The optimization algorithm is also able to react to changing traffic conditions and tunes the optimal speed profile to reach the destination within the specified arrival time. The optimization is applied on a conventional vehicle and also on a Series Hybrid Electric vehicle (SHEV). The fuel consumption optimization algorithm uses data obtained from quasi-static simulations. The algorithm is based on Dynamic Programming and divides the fuel consumption optimization problem into two parts. The first part of the optimization process reduces the fuel consumption according to foreseeable traffic conditions. It calculates an average speed profile that tries to avoid, if possible, the traffic jams on the road. Traffic jams that delay drivers result in higher vehicle speed to make up for lost time. A higher speed of the vehicle within an already defined time scheme increases fuel consumption. The second part of the optimization process is in charge of calculating the optimal speed profile according to the road slope and the remaining travel time. The optimization tunes the fuel consumption and travel time relevancies by using two penalty factors. Although the optimization results depend on the road slope and the travel time, the optimal speed profile produces improvements of 8.9% on the fuel consumption of the conventional car and of 2.8% on the spent energy of the hybrid vehicle when compared with a constant speed profile. The two parts of the optimization process are combined during the Real-Time execution of the algorithm. The average speed profile calculated by the optimization according to the traffic conditions provides values for the two penalty factors utilized by the second part of the optimization process. Although the savings depend on the road slope, traffic conditions, vehicle features, and the remaining travel time, simulations show that this joint optimization process can improve the energy consumption of the two vehicles types by more than 6%.