81690 research outputs found
Sort by
大学英語教員が認識する授業不安に関する実証的研究 : 影響要因と対処過程を視野に入れた混合研究法による分析 [全文の要約]
No full textこの博士論文全文の閲覧方法については、以下のサイトをご参照ください。https://www.lib.hokudai.ac.jp/dissertations/copy-guides
地方のスタートアップ・エコシステムの再検討 : 動機づけ理論に基づく起業家の多様性に関する実証研究 [全文の要約]
No full textこの博士論文全文の閲覧方法については、以下のサイトをご参照ください。https://www.lib.hokudai.ac.jp/dissertations/copy-guides
硫酸環境中におけるチタンの腐食に及ぼす金属カチオンの影響 [全文の要約]
No full textこの博士論文全文の閲覧方法については、以下のサイトをご参照ください。https://www.lib.hokudai.ac.jp/dissertations/copy-guides
Influence of Modeling Choices on Nonlinear-Dynamic Behavior of a Single-Layer Reticulated Dome
No full textNonlinear time history response analysis is a reliable approach for examining the large deformation of steel structures, particularly for the next generation structural design. The impact of modeling choices on the nonlinear response of steel structures has garnered significant attention in earthquake-prone regions. While the inertia force commonly retain the initial condition, the modeling of restoring and damping forces play a critical role in capturing nonlinear-behavior of steel structures. Fiber elements are frequently used to represent the nonlinearity of restoring forces because they can trace the propagation of plasticity along the length of structural members. However, the sensitivity of computational results of highly nonlinear-behavior, such as global buckling, local buckling and curvature distribution -- depends heavily on the discretization schemes and element formulations. Observation of damping ratios by vibration tests have provided strong evidence that these ratios remain relatively constant across relevant vibrations modes. However, since the vibration frequencies of inelastic system differ from those of elastic system, the use of damping models based on initial stiffness can result in unintended spurious damping forces. This study examines the influence of modeling choices on the dynamic response of a reticulated dome, analyzed by nonlinear time history response analysis using the platform OpenSees.
A reticulated dome with a span of 40.0 m and a rise of 13.3 m supported entirely by pinned connections, was selected as a representative model of spatial structures for seismic risk estimation. The dome was designed to remain stable under dead load and sustain insignificant and moderate damage states when subjected to Taft ground motions, with peak accelerations scaled to 0.45g and 0.7g based on a previous seismic risk study. Round hollow structural sections (HSS) were used for the individual members of the reticulated dome. Fiber elements based on force- and displacement-based formulations were validated against physical test results of round HSS members with slenderness ratios ranging from 40 to 74, which were commonly used in the dome. The validation considered models with 2, 4, 8, 10 and 30 segments. The displacement-based fiber element with 8 segments demonstrated high-fidelity in predicting the global response of the dome members in the reticulated dome with high-fidelity, accounting for plastic curvature induced by global buckling within the contraction range of -0.5% to 0.5%.
The force- and displacement-based fiber elements with 2, 4, 8 and 10 segments were utilized to investigate the impact of the member discretization on the dynamic response of the inelastic dome. The displacement-based fiber element with 8 segments demonstrated the highest accuracy in capturing both the global system and component-level behavior of the dome. The inherent damping of the reticulated dome was examined using three models: the initial-stiffness-proportional Rayleigh model, the tangent-stiffness-proportional Rayleigh model, and the modal-damping model. For the study of the damping models, the Taft ground motion and a suite of 28 pairs of ground motions selected by Table A-6C in (FEMA P695 2009), scaled to 1.3 times the design spectrum specified in Japanese Standard Law with a 10% probability of exceedance in 50 years, was employed. For all ground motion cases, at least 95% of the individual members exhibited contraction and elongation within ±0.5%. This result affirmed the reliability of the computational outcomes, as they aligned with the validated behavior of the members. A representative damping ratio for the dome was calculated using the modal-damping model incorporating a step-by-step update based on elastic-plastic behavior. The representative damping ratios in all three directions for the initial-stiffness-proportional damping were 1.26 times the target damping ratio, while those for tangent-stiffness-proportional Rayleigh damping and modal-damping model were closely matched the target value. Among the damping models, the tangent-stiffness-proportional Rayleigh model produced the largest 84th percentile displacement, whereas the initial-stiffness-proportional
Rayleigh model yielded the smallest. When the tangent-stiffness-proportional Rayleigh model was used, it produced the most severe damage evaluations compared to the initial-stiffness-proportional Rayleigh model and the modal damping model. The initial-stiffness-proportional Rayleigh model has a potential to yield less unconservative evaluation of damage states for spatial structures compared to the other two models. The modal-damping model is recommended for the seismic risk analysis to mitigate the effects of the spurious damping forces and ensure more accurate assessment for the reticulated domes.
Subsequently, to capture local buckling induced strength and stiffness deterioration in a steel concentrically braced frame (CBFs) calibration and extension of material model proposed by Suzuki and Lignos termed the SL Model was conducted. The SL Model was integrated into OpenSees to reproduce the hysteretic behavior of steel braces with round-HSS. The parameters for the SL Model were calibrated through regression analysis on round-HSS stud column data. The calibrated SL model was validated against round-HSS braces with slenderness ratios ranging from 16 to 103, and diameter-to-thickness ratios ranging from 16 to 63. The calibrated SL material model is supposed to be implemented in an 8-story chevron braced frame. In this frame, the lateral loads were primarily resisted by chevron braces, while the beams were designed as simply supported and proportioned to remain elastic under tension forces and brace compression, in accordance with the US Provisions
不均一系触媒設計のための触媒インフォマティクス・パイプラインの開発
No full textCatalysts are essential in modern chemistry and industry, enhancing reaction efficiency and enabling processes under milder conditions. They are widely used in applications such as fuel refining, pharmaceuticals, and energy and other raw materials production. Recent advances in Catalyst Design have started adopting data-driven methodologies, big data analysis, and high-throughput experimentation to accelerate the discovery of efficient catalyst materials. However, the integration of Informatics in Catalyst Design is still in an early stage, with challenges in establishing general workflows and representing catalysts as input descriptor variables in predictive models.
This thesis proposes a systematic workflow for screening catalyst candidates by designing and selecting compositional descriptor variables for Multiple Linear, Random Forest, and Support-Vector regression models. The workflow incorporates descriptor creation and engineering to represent catalysts, proposes the MonteCat algorithm as an efficient descriptor selection strategy to select for optimal descriptor variables, incorporates predictive modeling to evaluate catalyst performance for the Water-Gas Shift (WGS) and Oxidative Coupling of Methane (OCM) reactions and proposes potential catalyst candidates with experimental validation for low Temperature OCM catalysts.
Chapter 1 introduces the research context through a literature review, highlighting the challenges in WGS catalysts and the potential of OCM catalysts, along with objectives and the structure of the thesis. In summary, a workflow centered on constructing regression models from literature and experimental data to predict catalyst performance is proposed. Chapter 2 describes the methodology followed in further detail, namely the regression models explored, descriptor design to represent catalysts in predictive models, descriptor selection methodologies and the model validation strategy.
Chapter 3 explores catalyst representation for the Water-Gas Shift (WGS) reaction using a data set from reported literature. Two distinct catalyst representation schemes are explored: one-hot encoding and continuous compositional descriptors calculated from the catalyst compositions. This information, along with reaction conditions are used to construct regression models to predict catalyst performance and search for high conversion catalyst candidates through an inverse prediction.
Chapter 4 introduces the MonteCat algorithm, an optimization strategy inspired by Metropolis-Hastings and Basin Hopping algorithms, to enhance feature selection. MonteCat is used to construct descriptor subsets from a bank of over 3,000 descriptors, resulting in high-performing predictive models for the OCM reaction. Chapter 5 applies the MonteCat algorithm to design and experimentally validate OCM catalyst candidates. All tested materials demonstrated improved catalytic performance compared to the bare support.
Chapter 6 contains the general conclusions, which comprise the establishment of a workflow for catalyst performance prediction and a descriptor search algorithm that successfully proposed and validated catalyst materials for the OCM reaction. Future research should focus on integrating advanced descriptors with reactivity information to further enhance model interpretability and performance
