Développement de logiciel | EPFL Graph Search

Le développement de logiciel consiste à étudier, concevoir, construire, transformer, mettre au point, maintenir et améliorer des logiciels. Ce travail est effectué par les employés d'éditeurs de logiciels, des entreprises de services du numérique (ESN), des travailleurs indépendants (freelance) et des membres de la communauté du logiciel libre. Un logiciel est créé petit à petit par une équipe d'ingénieurs conformément à un cahier des charges établi par un client demandeur ou une équipe interne. Le logiciel est décomposé en différents modules et un chef de projet, ou architecte, se charge de la cohérence de l'ensemble. Différentes activités permettent de prendre connaissance des attentes de l'usager, créer un modèle théorique du logiciel, qui servira de plan de construction, puis construire le logiciel, contrôler son bon fonctionnement et son adéquation au besoin. La planification et la répartition des travaux permet d'anticiper le délai et le coût de fabrication. Le logiciel est

Influence of residual stresses on the buckling capacity of axially loaded steel columns

Gabriele Falconi

Over the past 60-80 years, the design methodology of steel structures comprising conventional steel profiles, such as HEA, HEB or HEM, has remained unchanged. However, during the 21st century, production methods have improved considerably. The aim of this study is to develop the basis for evaluating and refining the design methodology of steel members by considering realistic residual stress distributions generated during the fabrication process. Out of the numerous production methods, in this study we focus on hot rolled wide flange cross sections which are the most commonly used profiles in construction. Design specifications vary from one country to another. A comparison of the Swiss, European, Canadian, and Japanese approaches would help towards understanding the factors influencing the sizing of steel profiles. Simplified methods based on analytical and semi-experimental models have been developed over time to help engineers consider the phenomenon of lateral buckling by ensuring a sufficient degree of safety. Considering that the above design specifications have been developed based on experimental research of steel profiles and materials characteristic of old practices, this study investigates potential updates in the current design practices. To succeed in this challenge, it is imperative that the residual stress distributions of current cross-sectional and material characteristics are explained. Several studies throughout the past years have tried to focus on formulating simplified models that consider residual stress distributions. The accuracy of these models is investigated and recommendations are provided based on the best fits to measured residual stress distributions. These have explicitly been collected from nearly 10 experimental programs leading to more than 30 measured distributions with 𝐻/𝐵 (where 𝐻 is the height of the wide flange section and 𝐵 is the width) ranging from 0.95 to 2.22 (where 𝑡𝑓 is the thickness of the flange) ranging from 6.30 to 40mm. To supplement the knowledge regarding the residual stress distributions of current material specifications and to widen the cross-sectional range of available distributions, five additional wide flange residual stress distribution measurements (i.e., HEA160, HEM500, IPE120, IPE200 and IPE360) are implemented for the sake of this study. These distributions have been precisely determined through destructive laboratory measurements (i.e., cutting method) that are thoroughly explained in this thesis. In the advent of new technology and software development, structural element simulation can be performed in a thorough and precise manner. With the aid continuum finite element analysis (CFEA), the influence of residual stresses on lateral buckling of columns under axial load will be investigated. Eight cross-sectional profiles (i.e., HEA160, HEB500, HEM300, HEM500, IPE120, IPE200, IPE 360 and IPE400) are investigated. The CFEA consider realistic residual stress distributions that have been conducted by RESSLab, EPFL. The ultimate compressive load of these profiles considering varying buckling lengths (𝜆̅𝑘 ranging from 0.49 to 1.93, where 𝜆̅𝑘 is the normalised buckling length) are determined and compared with available design specifications.

2021

RoclET– Refactoring OCL Expressions by Transformations

Leander Eyer, Cédric Jeanneret, Slavisa Markovic

The role of UML models in software development has changed considerably over the last years. While UML was used in its early days mostly as a notation for sketching ideas, developers more and more recognize now the value of a UML model as a formal document that can be processed by tools, e.g. in order to generate code and documentation. Precise software models, however, can usually not be expressed by the pure diagrammatic elements of UML. Instead, they are best captured by constraints written in OCL. Despite the primary importance of constraints within precise modeling, the tool support for OCL is still in a rudimentary stage and there is much room for improvements. In this paper, we describe the architecture and the functionality of our own OCL tool called RoclET. Currently, our tool supports the parsing of OCL assertions (invariants, pre-/post-conditions), their evaluation in given system snapshots (object diagrams) and refactoring, i.e. the automatic propagation of changes made in the underlying class diagram across annotated OCL constraints. RoclET is highly customizable and can handle different OCL dialects. Its functionality is mainly implemented in form of QVT transformation rules that can be adapted by the user. RoclET is realized in form of an Eclipse plugin.

2006

Multi-object spectroscopic operations with the Sloan Digital Sky Survey V

Ricardo Jorge Araùjo Pereira, Loïc Grossen, Luzius Gregor Kronig

The Sloan Digital Sky Survey V (SDSS-V) is an all-sky spectroscopic survey of more than 6 million objects, designed to decode the history of the Milky Way, reveal the inner workings of stars, investigate the origin of solar systems, and track the growth of supermassive black holes across the Universe. SDSS-V presents significant innovations in both hardware and software, chiefly due to the introduction of a robotic Focal Plane System (FPS) that replaces plug-plate operations. This new mode of operations introduces new challenges with respect to target scheduling, fibre robot positioner reconfiguration optimisation, telescope guiding, observer interfaces, and observatory operations. During normal operations SDSS-V will observe a new field every 15 minutes. For each field requiring a new telescope pointing the FPS will reconfigure 500 robotic fibre positioners with feedback from an external Field Viewing Camera (FVC) in less than two minutes. Six CCD cameras mounted on the FPS will be used to automatically acquire the field and maintain the necessary guiding accuracy. These strict requirements highlight the need for streamlined operations software and procedures to minimise the time spent during FPS reconfigurations. We describe the overall design and implementation of the SDSS-V survey operations, with special emphasis on software development, conventions, and observing procedures. While specific to SDSS-V, the solutions we describe can be readily applied to other astronomical surveys and are of special interest given the rapid increase in projects employing robotic fibre positioners.

SPIE-INT SOC OPTICAL ENGINEERING2020

Influence of residual stresses on the buckling capacity of axially loaded steel columns

Gabriele Falconi

2021