京都大学 / 0048 / 新制・課程博士 / 博士(工学) / 甲第21091号 / 工博第4455号 / 新制||工||1692(附属図書館) / 京都大学大学院工学研究科都市社会工学専攻 / (主査)教授 五十嵐 晃, 教授 八木 知己, 准教授 古川 愛子 / 学位規則第4条第1項該当 / Doctor of Philosophy (Engineering) / Kyoto University / DFAM

Este trabalho tem como principal objetivo o desenvolvimento de um método numérico para simulação das grandes escalas de escoamentos turbulentos tridimensionais utilizando uma modelagem de turbulência baseada em filtragem temporal (denominada TLES - Temporal Large Eddy Simulation). O método desenvolvido combina discretizações temporais com ordem de mínima precisão 2 (Adams-Bashforth, QUICK, Runge-Kutta), um método de projeção de ordem 2, com discretizações espaciais também de ordem 2 obtidas pelo método de volumes finitos. Esta metodologia foi empregada na simulação de problemas teste turbulentos como o canal e a cavidade impulsionada, sendo este último resultado simulado pela primeira vez com modelagem TLES. Os resultados mostram uma excelente concordância quando comparado com resultados de simulações diretas (DNS) e dados experimentais, superando resultados clássicos obtidos com formulação LES com filtragem espacial. / The main objective of this work is to develop a numerical method for large eddy simulation of tridimensional turbulent flows using a model based on temporal filtering (TLES - Temporal Large Eddy Simulation). The developed method combines at least 2nd order temporal discretizations (Adams-Bashforth, QUICK, Runge-Kutta), a 2nd order projection method, and 2nd order spatial discretizations obtained by the finite volume method. This methodology was employed to the simulation of turbulent benchmark problems such as channel and lid-driven cavity flows. The latter is simulated for the first time using a TLES turbulence modelling. Results show excellent agreement when compared to Direct Numerical Simulations (DNS) and experimental data, with better results than classical results produced by standard LES formulation with spatial filtering.

Die zeitliche Filterung von sensorischem Input ist entscheidend für das Erkennen vieler Stimuli. Auditorische Neurone führen dazu mehrere Verarbeitungsschritte und Signaltransformationen durch, u.a. durch zeitliche Integration, zeitliche Auflösung und Selektion eines zeitlichen Merkmals. Um zu testen ob zeitliche Filterung von Stimuluseigenschaften (Intensität) oder Kontext (Temperatur) abhängt untersuchte ich Neurone in der Hörbahn der Wanderheuschrecke. Zuerst untersuchte ich zeitliche Integration in Rezeptoren und Interneuronen. Zeitverlauf und Ausmaß der Integration waren Neuronen-spezifisch. Während periphere Neurone die akustische Energie integrierten, unterschied sich die zeitliche Integration der Interneuronentypen stark, was eine spezifische zeitliche Filterung ermöglicht. Die Analyse postsynaptischer Potentiale deckte presynaptische und intrinsische Mechanismen der Integration auf, was darauf hindeutet, dass Unterschiede zwischen Neuronen wahrscheinlich auf Typ-spezifischer Verarbeitung beruhen. Zweitens erforschte ich die neuronale Antwort auf den zweiten Stimulus in einem Stimuluspaar mit einem Interstimulus-Intervall von wenigen Millisekunden. Die Veränderung der Antwort auf den zweiten im Vergleich zum ersten Stimulus zeigt den Effekt von akuter, kurzfristiger Adaptation und ist ein Maß für die maximale zeitliche Auflösung. In der sensorischen Peripherie trat moderate Adaptation auf, deren Einfluss exponentiell abfiel. Viele Interneurone zeigten dagegen nicht-lineare Effekte, wie die Unterdrückung oder Verstärkung der Antwort auf den zweiten Stimulus. Drittens testete ich den Effekt von Temperatur auf zeitliche Filterung. Die Selektivität von Interneuronen für zeitliche Stimulusmerkmale wurde bei wechselnden Temperaturen untersucht. Mit steigender Temperatur präferierten Neurone ein zeitlich komprimiertes Merkmal. Diese temperaturabhängige Veränderung könnte zur Temperatur-Kopplung von Sender und Empfänger bei den wechselwarmen Heuschrecken beitragen. / Temporal filtering of sensory input is crucial for the recognition of many sensory stimuli. Auditory neurons perform various computations and signal transformations to accomplish temporal filtering of acoustic input, comprising temporal integration, temporal resolution and temporal feature selection. To test whether temporal filtering processes within a neuron type depend on stimulus features, such as intensity, and on context, such as temperature, I conducted neurophysiological recordings from neurons in the auditory pathway of migratory locusts. First, I examined temporal integration in receptors and interneurons. The time course and extent of integration of subthreshold acoustic stimuli were neuronspecific. While peripheral sensory neurons acted as energy integrators, interneurons showed different temporal integration profiles, enabling neuron-specific temporal filtering. The analysis of postsynaptic potentials elucidated implemented mechanisms, suggesting that temporal integration is based on neuron-specific presynaptic and neuron-intrinsic computations. Second, I studied the response recovery of receptors and interneurons to the second stimulus in a stimulus pair, separated by a few milliseconds. This revealed the effect of acute, short-term adaptation and thus indicated the maximal temporal resolution of these neurons. In the sensory periphery response recovery was shaped by moderate adaptation and an exponential recovery. In many interneurons non-linear effects occurred, comprising a suppression of the response to the second stimulus and a response gain. Third, I tested the effect of temperature on temporal filtering. Temporal feature selectivity of interneurons was examined at cold and warm temperatures. With increasing temperature, the neurons preferred a temporally compressed feature. Temperature-dependent changes in temporal feature selectivity might thus contribute to temperature coupling of the sender and the receiver of the poikilothermic grasshoppers.

Este trabalho tem como principal objetivo o desenvolvimento de um método numérico para simulação das grandes escalas de escoamentos turbulentos tridimensionais utilizando uma modelagem de turbulência baseada em filtragem temporal (denominada TLES - Temporal Large Eddy Simulation). O método desenvolvido combina discretizações temporais com ordem de mínima precisão 2 (Adams-Bashforth, QUICK, Runge-Kutta), um método de projeção de ordem 2, com discretizações espaciais também de ordem 2 obtidas pelo método de volumes finitos. Esta metodologia foi empregada na simulação de problemas teste turbulentos como o canal e a cavidade impulsionada, sendo este último resultado simulado pela primeira vez com modelagem TLES. Os resultados mostram uma excelente concordância quando comparado com resultados de simulações diretas (DNS) e dados experimentais, superando resultados clássicos obtidos com formulação LES com filtragem espacial. / The main objective of this work is to develop a numerical method for large eddy simulation of tridimensional turbulent flows using a model based on temporal filtering (TLES - Temporal Large Eddy Simulation). The developed method combines at least 2nd order temporal discretizations (Adams-Bashforth, QUICK, Runge-Kutta), a 2nd order projection method, and 2nd order spatial discretizations obtained by the finite volume method. This methodology was employed to the simulation of turbulent benchmark problems such as channel and lid-driven cavity flows. The latter is simulated for the first time using a TLES turbulence modelling. Results show excellent agreement when compared to Direct Numerical Simulations (DNS) and experimental data, with better results than classical results produced by standard LES formulation with spatial filtering.