In this thesis biomimetic design is applied to the redesign of a PEM fuel cell flow field. A number of designs inspired by biological phenomena were developed to address the problem of attaining a uniform current density distribution across a PEM fuel cell. These designs are evaluated using a numerical model. One design, inspired by Murray’s law of branching in plants and animals, is further evaluated using and a physical model and comparing it to a commercial triple serpentine flow field. Improvements in pressure drop were seen for the Murray’s law inspired flow field, however, it was found to be prone to flooding. If this flow field design were to be applied to high temperature membrane materials, materials that can operate above 100 °C where water is always in the vapor state, the mass transfer and reduced pressure drop advantages of the Murray flow field could be fully achieved.

Water and R-134a sprays as they impinge on the flat endplate of a circle are studied experimentally. In order to optimize water and R-134a sprays cooling efficiency, a detailed characterization and understanding of the spray formation is essentially needed. The effects of the jet exit velocity and Weber number on spray segregation are investigated. An optical image system was used to quantify the droplet size and distribution. LDV measurements were used to characterize the local velocity and velocity fluctuation distribution from a commercial available nozzle in both axial and radial directions. It is found in the water spray that local mean droplet diameter (SMD) decreases as jet exit velocity increases and as jet proceeds further downstream as well. Furthermore, the SMD and radial velocity are found to be the largest at the outer edges of the water spray. In contrast, the radial velocity is found to be the smallest at the outer edges of the R-134a spray. The SMD and radial velocity continuously decrease across both the water spray and R-134a spray toward the jet axis; while the corresponding axial velocity is the maximum there. Moreover, the R-134a spray jet heat transfer in non-boiling regime was shown to be dependent on the velocity of the impinging jets in terms of Weber number and other related parameters which are in good agreement with those of previous studies.

Thanks to the rapid development of variable production patterns of industries, many kinds of chemicals are used. But the chemicals usually come with dangers of causing fires, explosions, and harming people. In the past decade in Taiwan, these chemicals caused many serious industrial disasters. They happened not only in conventional industries but also in semiconductor and chemical industries. And most of them happened due to the leaking of flammable or toxic gases. In the situation of the outbreak of fires, explosions may occur, and they will generally bring about heat radiation, explosive pressure, and energy releasing. And all of these often harm the workers and the environment, and also bring great loss for the factories. In order to prevent the disasters, except for improving the protection and safety equipment, it¡¦s more important to realize how to use effective ways to reduce them. If the gas pipes pass through a densely populated area, when some toxic gas leaks, it will cause fatal dangers which result from the mixing of gas diffusion and air, the flow process of the mixture, concentration of CH4, and temperature distribution, or explosions. After that, some toxic gas with poisonous substances will be released, and it will turn out to be horrible consequences that are beyond our imagination. So it¡¦s really important to do research on gas leaking and gas diffusion. If gas-leaking simulation is applied on the analysis of the leaking of flammable gas and gas diffusion, it¡¦s much more possible to protect the workers from being hurt, keep public safety, and reduce the loss on the wealth of the society. The thesis focuses on building various hazard patterns of gas leaking, gas explosions, and chemicals, etc. From all of these, the initial conditions and the degrees of dangers will be revealed. In the thesis, numerical simulation is used to analyze the density, pressure, speed of the leaking gas and all the distributions in the flow field. The major analysis is about the effects of parameter and the display of concentration distribution, and hazard range.

Direct Methanol Fuel Cells (DMFCs) are electrochemical energy conversion devices that convert chemical energy into electrical energy. The bipolar plate component of the DMFC is required to be fluid impermeable to prevent fuel leakage and electrically conductive to collect the electrons produced within the cell. Graphite possesses the properties of high electrical conductivity, low weight and resistance to corrosion that make it an attractive material for bipolar plates. However, the poor mechanical properties of graphite lead to prohibitive machining costs and increased production times. The objective of this research is to develop an indirect laser sintering (LS) process, involving the laser sintering of graphite powders mixed with a phenolic resin binder which offers the advantage of complex part production and testing of prototype bipolar plates in short times. Due to the nature of the indirect LS process, the as-produced (green part) plates are porous and possess low electrical conductivities (< 0.1 S.cm-1). This research describes a viable method to rapidly fabricate and test multiple graphite bipolar plate designs using indirect LS. This process involved identifying and selecting suitable graphite powder and binder systems based on their thermal and electrical properties and developing a post process heat treatment method for achieving electrical conductivity of 250 S/cm for LS graphite parts which exceeds the DOE target of 100 S/cm for bipolar plate materials. The post processing also covered a method of infiltration using cyanoacrylate which was capable of rendering porous brown parts fluid impermeable and suitable for use in DMFCs. The cyanoacrylate infiltrated LS graphite parts were characterized for flexural strength and electrical and thermal conductivities and bipolar plates were made and evaluated in a DMFC test stand. Various flow field designs including plates with varying channel and rib widths and triangular, elliptical and rectangular flow field cross sections were fabricated using indirect LS and their respective polarization curves were compared to commercially machined graphite plates. The fuel cell tests show the improvement in mass transport performance could be due to improved methanol distribution and water removal characteristics of triangular and elliptical cross sectional channels over rectangular channels of equivalent dimensions. / text

The performance of a new cathode flow field plate located on a PEM fuel cell was compared to an industry standard and optimal serpentine design provided from literature. Results were successfully collected through a fuel cell module integrated with the 3D computational fluid dynamics package ANSYS Fluent. Contour plots showing a cathode catalyst layer comparison of local current density, oxygen molar concentrations, water content, and the pressure inside of the flow channels were compared with both PEM fuel cell configurations. The new flow field plate/pattern was shown to distribute more mass species of oxygen, more evenly, to the reaction site given the same boundary conditions, thus contributing to more ideal local current density. The net-power was determined for both fuel cells which included the pump work-in and power-out from each fuel cell. The new flow field plate was shown, through computational power performance results, to outperform the conventional flow pattern by up to 2.4% when excluding the effects of pump work, and still upheld a positive gain when factoring in this value. With an additional 18 corners for improved water management due to the effects of wall adhesion, the new bipolar plate was proven to become a new competitor in PEM fuel cell technology. Furthermore, this thesis gives further insight on PEMFC digital prototyping.

Identifying marine pollutant sources is essential in order to assess, contain and minimize their risk. We propose a Lagrangian Particle Tracking algorithm (LPT) to study the transport of passive tracers continuously released from fixed and moving sources and to identify their source in a backward mode. The LPT is designed to operate with uncertain flow fi elds, described by an ensemble of realizations of the sea currents. Starting from a region of high probability, re- verse tracking is used to generate inverse maps. A probability-weighted distance between the resulting inverse maps and the source trajectory is then minimized to identify the likely source of pollution. We conduct realistic simulations to demonstrate the efficiency of the proposed algorithm in the Mediterranean Sea using ocean data available from Copernicus Marine Environment Monitoring Services. Passive tracers are released along the path of a ship and propagated with an ensemble of flow fi elds forward in time to generate a probability map, which is then used for the inverse problem of source identi fication. Our experiments suggest that the algorithm is able to efficiently capture the release time and source, with some test cases successfully pinpointing the release time and source up to two weeks back in time.

京都大学 / 0048 / 新制・課程博士 / 博士(工学) / 甲第21366号 / 工博第4525号 / 新制||工||1705(附属図書館) / 京都大学大学院工学研究科機械理工学専攻 / (主査)教授 花崎 秀史, 教授 黒瀬 良一, 教授 稲室 隆二 / 学位規則第4条第1項該当 / Doctor of Philosophy (Engineering) / Kyoto University / DGAM

Sediment material around the base of a bridge pier is moved by the flow velocity and associated turbulence. This phenomenon is generally termed as local scour and can lead to undermining the structure and increase its possibility of failure. Numerous factors can affect bridge pier scour and they have been investigated for decades. Debris jams, one of these factors, could significantly contribute to bridge failure as some field examples and experimental investigations pointed out. Woody debris accumulation on the front of either single or multiple bridge piers can result in deeper pier scour and extra load exerted on the pier. Several studies have already investigated the influence of woody debris on pier scour in terms of static woody debris. In addition, HEC-18 (2012) also proposed a design code to estimate scour depth in the presence of woody debris jam. However, in these studies, the woody debris jam was considered to be static, whereas a woody debris jam accumulates piece by piece, growing to a debris jam with a shape most akin to a half-cone, and then may even eventually break up and be carried in pieces downstream. Therefore, this research investigated the evolution of the loading onto and scouring around a bridge pier in the presence of dynamic debris jams. In this study, the temporal evolution of the bridge pier scours was monitored during the development of dynamic debris jams. Experimental modeling was conducted to explore the influence of dynamic debris jam on bridge pier scour using a scale of 30 by employing both dowels and seedling trees. It was found that the dynamic debris jam of dowels could last 10-20 minutes and reach a critical size, then fail and subsequently reform. In addition, the first debris jam had an obvious influence on scour depth which correlated to the blockage generated by the debris jam; however, the influence of the subsequent debris jam depended on its size compared to the previously formed one. For the dynamic debris jam using seedling trees, the debris jam lasted for a longer time once it formed, and it could lead to twice the maximum scour depth compared to that generated in the absence of the debris jam, which is the same with dowels debris jam. In addition, the hydraulic head induced by the debris jam was correlated to the blockage of the debris jam and the flow Froude number irrespective of whether the dynamic debris jam was made of dowels or seedling trees. Additionally, blank control tests in the absence of a debris jam were used along with previous data gleaned from the literature to develop and test new multigene genetic programming (MGGP) models for the temporal evolution of scour. The MGGP model, using the non-dimensional variables from the empirical equations, can reach a better accuracy than the empirical equations, which indicates the ability of the model to optimize the empirical equations. The temporal evolution of load exerted onto the bridge pier with a dynamic debris jam was also measured. Experimental tests were performed to investigate the additional debris jam drag force exerted onto the bridge pier using both dowels and seedling trees in the presence of a fixed flume bed. Likewise, the dynamic debris jam of dowels lasted for about 10-20 mins, while those formed by the seedling trees, once formed, could last over 50 mins. The investigation demonstrated that the drag coefficient of the seedling trees jam was higher than that of the dowels jam. More importantly, a spike in the drag force was also observed irrespective of whether the jams were formed by dowels or seedling trees. Detailed investigation of the flow field around the debris jam and pier provided insight into the mechanics of debris jams. Three half-cone-shaped debris jams of the same dimensions were designed and built. The three jams were fabricated using: a) 20 cm long dowels, b) 30 cm long dowels, or c) a 3D printer. For each jam, four sections were measured using an Acoustic Doppler Velocimeter (ADV). The results indicated that the flow fields around the 20 cm length dowel jam and the 30 cm length dowel jam were similar. In addition, the section behind the pier and debris jam showed divided zones termed herein as the accelerated high-velocity zone, the high shear transition zone, and the wake dead zone. As for the drag coefficient, the 20 cm length dowels jam and 30 cm length dowels jam shared a very close magnitude of 1.7, but the drag coefficient of the 3D printer debris jam was only 0.88 which indicated the debris jam built by individual pieces behaved differently than the block jam.

Yes / The present work deals with the flow-induced multiple orientations and crystallization structure of polymer melts under a complex flow field. This complex flow field is characteristic of the consistent coupling of extensional “pulse” and closely followed shear flow in a narrow channel. Utilizing an ingenious combination of an advanced micro-injection device and long chain aliphatic polyamides (LCPA), the flow-induced crystallization morphology was well preserved for ex-situ synchrotron micro-focused wide angle X-ray scattering (μWAXS) as well as small angle X-ray scattering (SAXS). An inverted anisotropic crystallization structure was observed in two directions: perpendicular and parallel to the flow direction (FD). The novel anisotropic morphology implies the occurrence of wall slip and “global” fountain flow under the complex flow field. The mechanism of structure formation is elucidated in detail. The experimental results clearly indicate that the effect of extensional pulse on the polymer melt is restrained and further diminished due to either the transverse tumble of fountain flow or the rapid retraction of stretched high molecular weight tails. However, the residual shish-kebab structures in the core layer of the far-end of channel suggest that the effect of extensional pulse should be considered in the small-scaled geometries or under the high strain rate condition.

Rivers are considered as one of the most important surface water resources on the earth. During the time, most of the rivers on the earth experienced evolution and changes. River bends and confluences are one of the common cases in most rivers. There is a significant impact of the flow on the cross-sectional profile of river bends and confluences. Secondary currents are one of the important features that characterize flow in river bends and confluences. In such currents, fluid particles follow a helical path instead of moving nearly parallel to the axis of the channel. The local imbalance between the vertically varying centrifugal force and the cross-stream pressure gradient results in generating the secondary flow and raising a typical motion of the helical flow. Several studies, including experimental or mathematical, have been conducted to examine flow characteristics in curved open channels, river meanders, or confluences. In this research, the influence of secondary currents is studied on the elevation of water surface and the hydraulic structures in channel bends and confluences by employing a 3D OpenFOAM numerical model. The research implements a 3D OpenFOAM numerical model to simulate the horizontal distribution of the flow. In addition, the progress in unraveling and understanding the bend and confluent dynamics is discussed. The finite volume method in OpenFOAM software is used to simulate and examine the behavior of the secondary current. Thereafter, a comparison between the experimental data and a numerical model is conducted. Two sets of experimental data are used as the dataset for these two experiments are complete and validated; the data provided by Rozovskii (1961) for a sharply curved channel, and the dataset provided by Shumate (1998) for a confluent channel. Two solvers in OpenFOAM software were selected to solve the problem regarding the experiment: InterFoam and PisoFoam. InterFoam is a transient solver for incompressible flow that is used with open channel flow with Free Surface Model. PisoFoam is a transient solver for incompressible flow that is used with closed channel flow and Rigid-Lid Model. Various turbulence models (i.e., Standard k-ε, Realizable k-ε) are applied in the numerical model to assess the accuracy of turbulence models in predicting the behavior of the flow. The accuracies of various turbulence models are examined and discussed.