Choose an application

• Reference Manager
• EndNote
• RefWorks (Direct export to RefWorks)

At present, Eurocode 1 Part 1-7 presents different strategies to deal with the problem of robustness. However, for each of these strategies, the content is limited to the statement of general principles. Consequently, in recent years, several researches have been carried out to develop calculation and design tools for robustness. Thanks to these researches, many new tools and works are available to practitioners, but the lack of comprehensive approaches limits their scope and use.

In this work, approaches for handling the robustness problem in steel and composite structures are developed for the strategies presented in the Eurocode, and based on the tools currently available. For each of the approaches developed, the tools it requires to use, the limitations it presents and the analysis and verification steps to be followed are presented in detail. In addition, all this information is summarized in a global flowchart proposed to deal with the robustness problem.

This first task highlights several inconsistencies concerning the prescriptive tying method currently proposed by the Eurocode. Proposals for improvements have actually already been made recently.

Therefore, in a second step, this thesis focuses more specifically on this tying method. To assess the relevance of the rules currently proposed by the Eurocode regarding the tying method and the degree of robustness which these rules allow to reach, a study of a braced structure undergoing the loss of a column is carried out. In the context of this study, the beam-column joints of the structure first have the resistance specified by the tying method. Subsequently, different joint configurations are studied. By comparing the results obtained, several conclusions are drawn about the tying method. Also, this study shows that simple joints such as header plate connections have a too low degree of robustness, and that it is more advantageous to use semi-rigid/rigid joints such as flush end-plate connections.

robustness --- strategies --- tying method --- Ingénierie, informatique & technologie > Ingénierie civile

Choose an application

• Reference Manager
• EndNote
• RefWorks (Direct export to RefWorks)

In the early fties, Hodgkin and Huxley developed a model of the electrical activity of the neuron. Based on a simple RC circuit with non-linear conductances, they reproduced very well the electrical behavior of a squid neuron. Over the last fty years, thanks to the increase of experimental data and knowledge in neuroscience, scientists have extended the Hodgkin and Huxley's model to more complex neurons. But they have often increased the complexity which makes their models less robust. 
This thesis focuses on thalamic neurons. The thalamus is the relay-station for the sensory inputs travelling to the cortex. Depending on the state, the thalamic neurons exhibit two di erent ring patterns. During sleep, the neurons are bursting, which stops the information processing. During wakefulness, the neurons are spiking and the thalamus processes the inputs. In order to study diseases such as absence seizures in children, or to describe more precisely the thalamic behavior, a robust model of neuron activity switch is necessary. This robustness has to be maintained when the study is performed at the network level.

Recent evidences have highlighted the critical role of the slow dynamics of neuronal calcium currents in the switch from spiking to bursting. Inspired by this line of work, this thesis gathers conductance-based models of thalamic neuron in the literature. The major di erence between them is the presence of the slow kinetics of the calcium current. The rst contribution is their robustness comparison at the cellular level. Models which lack this slow dynamic are fragile when they are subjected to perturbation.
The second contribution is to show that this slow dynamic is necessary to reproduce the correct rhythmicity of the thalamus at the network level.

The conductance-based models are powerful tools to simulate a neuron with a great biophysical realism. However, they consist in high-dimensional non-linear di erential equations that lead to time-consuming simulations. Therefore, the second part of this thesis investigates simple, qualitative modeling of neuron and network activity. This type of model, called hybrid model, is more mathematical; it captures the subthreshold dynamics of the neuron through di erential equations and adds a reset rule to mimic the all-or-none nature of the spike. A hybrid model of a thalamic neuron has to be able to switch from spiking to bursting. Its robustness at the cellular level relies on its ability to mimic the slow dynamics of the calcium current without mathematical manipulation.
The third contribution of the thesis is to con rm this discussion with a network level analysis. It shows that previously available simple models of thalamocortical neurons such as the well-known Izhikevich models lack the slow dynamics, hence they generate pathological behaviors while connected within a circuit.

The key message is the comparison between two classes of thalamic neuron models. The rst class integrates the slow dynamics of the calcium current while the second class assumes that this dynamics is fast. This work shows that the rst class provides better results in terms of robustness. This demonstration is led at the cellular and network levels, for conductance-based models or reduced models. Therefore, the models belonging to this class are suitable for studies concerning the neuromodulation or the synaptic plasticity. Au début des ann ées cinquante, Hodgkin et Huxley ont développé un modèle de l'activité électrique neuronale. Bas é sur un simple circuit RC caract eris é par des conductances non-linéaires, ils ont réussi à reproduire de manière précise le comportement électrique d'un neurone de calamar. Durant ces dernières années, l'augmentation du nombre de données expérimentales disponibles, ainsi qu'une amélioration des connaissances dans le domaine de la neuroscience, ont permis aux scienti ques d' étendre le mod èle de Hodgkin et Huxley au cas de neurones plus complexes. Cependant, ces modèles augmentent également en complexité mathématique, ce qui les rend moins robustes.
Cette thèse se concentre uniquement sur les neurones du thalamus. Cette partie du cerveau est le
centre-relais des informations sensorielles voyageant vers le cortex. En fonction de leur état, les neurones du thalamus sont caractérisés par deux motifs de décharge. Durant le sommeil, les neurones "burstent", ce qui bloque le traitement de l'information. Durant la phase d'éveil, les neurones présentent un enchaînement régulier de pics qui permettent au thalamus de traiter l'information et de l'envoyer au cortex. Un model robuste décrivant ce changement d'activité est primordial afin de mieux comprendre certaines maladies telles que l'absence d'épilepsie ou de d'écrire plus précisément le comportement du thalamus.

De récentes études ont mis en évidence le rôle critique de la dynamique lente des courants calciques présents dans les neurones dans la transition entre les deux modes de décharge. Inspirée par ces recherches, cette thèse rassemble des modèles à conductances des neurones du thalamus présents dans la littérature. La différence majeure entre ces modèles réside dans l'intégration ou non de la cinétique lente des courants calciques. La première contribution de ce travail est la comparaison de leur robustesse à l'échelle cellulaire. Les modèles qui omettent cette dynamique lente sont fragiles lorsqu'ils sont soumis à des perturbations.
La deuxième contribution consiste à montrer que cette caractéristique des courants calciques est nécessaire pour reproduire le rythme d'une population de neurones du thalamus.

Les modèles à conductances sont des outils puissants pour simuler un neurone avec une bonne interpretation biophysique. Cependant, ils sont formés d'un grand nombre d'équations différentielles non-linéaires menant à des simulations couteuses en temps. Par cons équent, la deuxi ème partie de cette thèse s'oriente vers une modélisation plus simple et plus qualitative des neurones et de leur activit é en r éseau. Ce type de modèle, appelé modèle hybride, est plus mathématique ; il capture la dynamique du signal neuronal au travers une équation différentielle. Ensuite, une équation de remise à zéro, appel ée la r ègle du "reset", tient compte de la nature "tout ou rien" des pics présents dans le signal électrique. Un modèle hybride d'un neurone du thalamus doit être capable de reproduire la transition entre les deux modes de décharges. Sa robustesse à l'échelle cellulaire repose sur son aptitude à imiter la dynamique lente des courants calciques sans manipulation mathématique.
La troisième contribution de cette thèse est de confirmer cette hypothèse avec une analyse à l'échelle d'un r éseau de neurones. Cette étude prouve que les modèles plus simples des neurones du thalamus présents dans la litt érature, tels que les modèles d'Izhikevich, n'intègrent pas cette cin étique lente. Par conséquent, ils ne sont pas capables de reproduire l'activité rythmique du thalamus.

Pour résumer, cette thèse à pour but de comparer deux classes de modèles de neurones du thalamus. Une classe intègre la dynamique lente des courants calciques en opposition à l'autre classe qui assume que cette dynamique est rapide. Ce travail montre que la classe faisant l'hypothèse d'une dynamique lente donne des résultats favorables en terme de robustesse. Cette d émonstration est menée au niveau cellulaire et à l'échelle d'un réseau de neurones, tant pour des modèles a conductances que des modèles r éduits.

thalamus --- thalamic neuron model --- robustness --- calcium channels --- neuromodulation --- Ingénierie, informatique & technologie > Ingénierie électrique & électronique

Choose an application

• Reference Manager
• EndNote
• RefWorks (Direct export to RefWorks)

The request for robustness of structures is a recent topic aiming at ensuring the structural integrity of a structure in case of exceptional events. General design recommendations are provided in modern codes and standards but they are far from being satisfactory as it has been demonstrated that the respect of these recommendations do not necessarily allow to guarantee a sufficient robustness to a structure. It is the reason why research activities are still ongoing in this field in view of (i) mastering the response of structures when subjected to exceptional events and, on this basis, (ii) to derive rules for practitioners. This master thesis takes place in this context.

The objective of this master thesis is to study and apply methods for steel building structures allowing to reach an appropriate level of robustness. These research will be performed in the framework of an ongoing European RFCS project entitled "FailNoMore". 

In particular, a 3D steel structures, initially designed by a German design office for “classical” loading conditions, will be studied in view of (i) characterising its behaviour when subjected to a specific exceptional event, i.e. the loss of a supporting member, and of (ii) investigating the efficiency of existing design methods for robustness. At the end of the work, a critical analysis of the results obtained with the different methods will be provided in view of making recommendations for practice. 

These objectives will be achieved using numerical and analytical approaches. The numerical investigations will be realised using the FINELG software. 

This master thesis will show that the tying method, such as it is present in the Eurocode, does not make it possible to ensure a sufficient robustness in the case of the scenario considered of the loss of a column.
It is also shown that, following a numerical study, the weakness of the structure is in the connections. A structure with hinged connections is not robust when column loss is considered.
By using partially resistant connections, and by means of an innovative analytical approach, it is shown that the considered structure can be robust under the exceptional scenario of column loss by slightly modifying the properties of the structure.

Robustness --- Tying method --- Steel structure --- FailNoMore --- Ingénierie, informatique & technologie > Ingénierie civile