京都大学 / 新制・課程博士 / 博士(工学) / 甲第25279号 / 工博第5238号 / 京都大学大学院工学研究科機械理工学専攻 / (主査)教授 黒瀬 良一, 教授 長田 孝二, 教授 平山 朋子 / 学位規則第4条第1項該当 / Doctor of Philosophy (Engineering) / Kyoto University / DGAM

Thesis (MScEng (Mechanical and Mechatronic Engineering))--Stellenbosch University, 2008. / Aortic valve replacements are frequently performed during heart surgery. However, since this is quite a stressful procedure, many patients are turned down for medical reasons. Stented valves, designed and manufactured for percutaneous insertion, eliminate many of the risks involved in open-heart surgery, thus providing a solution to patients not deemed strong enough for open-chest aortic valve replacements. The aortic valve is a complex structure, and therefore numerical simulation is necessary to obtain flow and stress data to support the design of a prosthetic heart valve in the absence of viable physical measuring methods. To aid in the design of a prosthetic heart valve, various finite element valve models were created, and the fluid structure interaction (FSI) between the valves and the blood was simulated using commercial finite element software. The effect of the geometry of the leaflets on the haemodynamic behaviour over the cardiac cycle was investigated. It was found that leaflet dimensions should be chosen judiciously, because of their considerable effect on the stress distribution and performance of the valve. A simple leaflet geometry optimisation was done for a 20 mm and 26 mm valve, respectively, by means of existing geometry relationships found in the literature. Simulations were done to obtain the maximum leaflet attachment forces that can be used by a stent designer for fatigue loading, or to investigate the structural strength of the stent. These simulations were numerically validated. The effect of leaflet thickness and stiffness on resistance to opening, stress distribution and strain were investigated. Results showed that leaflet thickness has a greater effect on the performance of the valve than leaflet stiffness, and thereby validated the results of similar tests contained in the literature. After simulating over-, as well as under-dilation of a stented valve, it was found that problems associated with over-dilation can be minimised to a certain extent by increasing the coaptation1 region of the leaflets. A simple pulse duplicator was designed based on a four-element Windkessel model. The pulse duplicator was used to study the performance of the prototype valves by means of high-speed photography, the results of which were fed into one of the numerical finite element models and compared to real valve performance. Some of the prototype valves showed efficiencies of 88%.

Thesis (MScEng)--Stellenbosch University, 2012. / ENGLISH ABSTRACT: Bioprosthetic heart valves (BHVs) for transcatheter aortic valve implantation (TAVI) have been rapidly developing over the last decade since the first valve replacement using the TAVI technique. TAVI is a minimally invasive valve replacement procedure offering lifesaving treatment to patients who are denied open heart surgery. The biomedical engineering research group at Stellenbosch University designed a 19 mm balloon expandable BHV for TAVI in 2007/8 for testing in animal trials. In the current study the valve was enlarged to 23 mm and 26 mm diameters. A finite element analysis was performed to aid in the design of the stents. New stencils were designed and manufactured for the leaflets using Thubrikar‟s equations as a guide. The 23 mm valve was manufactured and successfully implanted into two sheep. Fluid structure interaction (FSI) simulations constitute a large portion of this thesis and are being recognized as an important tool in the design of BHVs. Furthermore, they provide insight into the interaction of the blood with the valve, the leaflet dynamics and valve hemodynamic performance. The complex material properties, pulsating flow, large deformations and coupling of the fluid and the physical structure make this one of the most complicated and difficult research areas within the body. The FSI simulations, of the current valve design, were performed using a commercial programme called MSC.Dytran. A validation study was performed using data collected from a cardiac pulse duplicator. The FSI model was validated using leaflet dynamics visualisation and transvalvular pressure gradient comparison. Further comparison studies were performed to determine the material model to be used and the effect of leaflet free edge length and valve diameter on valve performance. The results from the validation study correlated well, considering the limitations that were experienced. However, further research is required to achieve a thorough validation. The comparative studies indicated that the linear isotropic material model was the most stable material model which could be used to simulate the leaflet behaviour. The free edge length of the leaflet affects the leaflet dynamics but does not greatly hinder its performance. The hemodynamic performance of the valve improves with an increase in diameter and the leaflet dynamics perform well considering the increased surface area and length. Many limitations in the software prevented more accurate material models and flow initiation to be implemented. These limitations significantly restricted the research and confidence in the results. Further investigation regarding the implementation of FSI simulations of a heart valve using the commercial software is recommended. / AFRIKAANSE OPSOMMING: Bio-prostetiese hartkleppe (Bioprosthetic Heart Valves - BHVs) wat gebruik word vir transkateter aortaklep-inplantings (Transcatheter Aortic Valve Implantation - TAVI) het geweldig vinnige ontwikkeling getoon in die afgelope tien jaar sedert die eerste klepvervanging wat van die TAVI prosedure gebruik gemaak het. TAVI is ʼn minimaal indringende klepvervangingsprosedure wat lewensreddende behandeling bied aan pasiënte wat ope-hart sjirurgie geweier word. Die Biomediese Ingenieurswese Navorsingsgroep (BERG) by Stellenbosch Universiteit het in 2007/8 ʼn 19 mm ballon-uitsetbare BHV vir TAVI ontwerp vir eksperimente met diere, en hierdie tesis volg op die vorige projekte. In die huidige studie is die klep vergroot na 23 mm en 26 mm in deursnee. ʼn Eindige element analise is gedoen om by te dra tot die ontwerp van die rekspalke vir die klep. Nuwe stensils is ontwerp en vervaardig vir die klepsuile, deur gebruik te maak van Thubrikar se vergelykings. Die 23 mm klep is vervaardig en suksesvol in twee skape ingeplant. Vloeistruktuur interaksie (Fluid Structure Interaction (FSI)) simulasies vorm ‟n groot deel van die tesis en word gesien as ʼn noodsaaklike hulpmiddel in die ontwerp van BHVs. Die simulasies verskaf ook insig in die interaksie van die bloed met die klep, die klepsuil-dinamika en die klep se hemodinamiese werkverrigting. Die komplekse materiaal eienskappe, polsende vloei, grootskaalse vervorming, die verbinding van die vloeistof en die fisiese struktuur maak van hierdie een van die mees gekompliseerde voorwerpe om te simuleer. Die FSI simulasies van die huidige ontwerp, is uitgevoer deur van kommersiële sagteware, MSC.Dytran, gebruik te maak. ʼn Geldigheidstudie wat data gebruik het vanaf die hartklop-nabootser, is uitgevoer. Die FSI model word geverifieer deur klepsuil dinamika visualisering en ʼn vergelyking van die drukgradiënt gebruik te maak. Verdere vergelykende studies is uitgevoer om te bepaal watter materiaal model om te gebruik, asook die uitwerking van die klepsuil-vrye rand en klepdeursnee op die klep se werkverrigting. Die resultate van die studie korreleer goed, in ag genome die beperkings wat ervaar is. Verdere navorsing is egter nodig vir ʼn volledige geldigheidstudie. Vergelykende studies het getoon dat die liniêre isotropiese materiaalmodel die meer stabiele materiaalmodel is wat kan gebruik word om klepsuilgedrag te simuleer. Die vrye-rand lengte van die klepsuil affekteer die dinamika van die klepsuil, maar belemmer nie die werkverrigting grootliks nie. Die hemodinamiese werkverrigting van die klep verbeter met die toename in deursnee en die klepsuil-dinamika vertoon goed in ag genome die verhoogde oppervlak area en lengte. Die vele beperkings in die sagteware het die implementering van meer akkurate materiaalmodelle verhoed. Hierdie beperkings het ʼn verminderde vertroue in die resultate tot gevolg gehad. Verdere ondersoek rakende die implementering van die FSI simulasies van ʼn hartklep deur kommersieel beskikbare sagteware te gebruik, word aanbevel.

The response of submerged structures subject to loading by underwater blast waves is governed by complex interactions between the moving or deforming structure and the surrounding fluid and these phenomena need to be thoroughly understood in order to design structural components against underwater blast. This thesis has addressed the response of simple structural systems to blast loading in shallow or deep water environment. Analytical models have been developed to examine the one-dimensional response of both water-backed and air-backed submerged rigid plates, supported by linear springs and loaded by underwater shock waves. Cavitation phenomena as well as the effect of initial static fluid pressure are explicitly included in the models and their predictions were found in excellent agreement with detailed FE simulations. Then, a novel experimental apparatus has been developed, to reproduce controlled blast loading in initially pressurised liquids. It consists of a transparent water shock tube and allows observing the structural response as well as the propagation of cavitation fronts initiated by fluid-structure interaction in a blast event. This experimental technique was then employed to explore the one-dimensional response of monolithic plates, sandwich panels and double-walled structures subject to loading by underwater shock waves. The performed experiments provide great visual insight into the cavitation process and the experimental measurements were found to be in good agreement with analytical predictions and dynamic FE results. Finally, underwater blast loading of circular elastic plates has been investigated by theoretically modelling the main phenomena of dynamic plate deformation and fluid-structure interaction. In addition, underwater shock experiments have been performed on circular composite plates and the obtained measurements were found in good correlation with the corresponding analytical predictions. The validated analytical models were then used to determine the optimal designs of circular elastic plates which maximise the resistance to underwater blast.

Der mit flüssigem Natrium gefüllte Druckbehälter eines Präzessionsexperimentes unterliegt im Betrieb einer Vielzahl an Belastungen. Neben den aus der Rotation und Präzession resultierenden Fliehkräften und dem gyroskopischen Moment müssen ebenfalls die fertigungsbedingten Unwuchten sowie die Fluid-Struktur-Interaktion berücksichtigt werden. Darüber hinaus stellen die bei der Erwärmung bzw. Abkühlung auftretenden thermischen Spannungen eine wesentliche Beanspruchung dar. Es wird ein Algorithmus vorgestellt, der es ermöglicht, alle diese transienten und winkelabhängigen Lasten bei minimalem Rechenaufwand in den Berechnungsprozess einzubeziehen und die für den statischen und zyklischen Festigkeitsnachweis nach FKM-Richtlinie maßgeblichen Beanspruchungen zu identifizieren. Dies ermöglicht die vollflächige Berechnung des Auslastungsgrades in ANSYS Workbench.

In a proposed design for a concentrated solar power tower, sand is irradiated by solar energy and transfers its energy to another fluid stream by means of a finned tube heat exchanger. To maximize heat transfer and minimize potential damage to the heat exchanger, it is desired to have a very uniform flow through the heat exchanger. However, performing full scale flow tests can be expensive, impractical, and depending upon the specific quantities of interest, unsuitable for revealing the details of what it happening inside of the flow stream. Thus, the discrete element method has been used to simulate and study particulate flows. In this project, the flow of small glass beads through a square pyramid shaped hopper and a wedge shaped hopper were studied at the lab scale. These flows were also simulated using computers running two versions of discrete element modeling software – EDEM and LIGGGHTS. The simulated results were compared against the lab scale flows and against each other. They show that, in general, the discrete element method can be used to simulate lab scale particulate flows as long as certain material properties are well known, especially the friction properties of the material. The potential for increasing the accuracy of the simulations, such as using better material property data, non-uniform particle size distributions, and non-spherical particle shapes, as well as simulating heat transfer within a granular flow are also discussed.

Tails, fins, scales, and surface coatings are used by organisms for various tasks, including locomotion. Since millions of years of evolution have passed, we expect that the design of surface structures is optimal for the tasks of the organism. These structures serve as an inspiration in this thesis to identify new mechanisms for flow control. There are two general categories of fluid-structure-interaction mechanisms. The first is active interaction, where an organism actively moves parts of the body or its entire body in order to modify the surrounding flow field (e.g., birds flapping their wings). The second is passive interaction, where appendages or surface textures are not actively controlled by the organism and hence no energy is spent (e.g., feathers passively moving in the surrounding flow). Our aim is to find new passive mechanisms that interact with surrounding fluids in favourable ways; for example, to increase lift and to decrease drag. In the first part of this work, we investigate a simple model of an appendage (splitter plate) behind a bluff body (circular cylinder or sphere). If the plate is sufficiently short and there is a recirculation region behind the body, the straight position of the appendage becomes unstable, similar to how a straight vertical position of an inverted pendulum is unstable under gravity. We explain and characterize this instability using computations, experiments and a reduced-order model. The consequences of this instability are reorientation (turn) of the body and passive dispersion (drift with respect to the directionof the gravity). The observed mechanism could serve as a means to enhance locomotion and dispersion for various motile animals and non-motile seeds. In the second part of this thesis, we look into effective models of porous and poroelastic materials. We use the method of homogenization via multi-scale expansion to model a poroelastic medium with a continuum field. In particular, we derive boundary conditions for the velocity and the pressure at the interface between the free fluid and the porous or poroelastic material. The results obtained using the derived boundary conditions are then validated with respect to direct numerical simulations (DNS) in both two-dimensional and three-dimensional settings. The continuum model – coupled with the necessary boundary conditions – gives accurate predictions for both the flow field and the displacement field when compared to DNS. / Många djur använder sig av fjäll, päls, hår eller fjädrar för att öka sin förmåga att förflytta sig i luft eller vatten. Eftersom djuren har genomgått miljontals år av evolution, kan man förvänta sig att ytstrukturernas form är optimala för organismens uppgifter. Dessa strukturer tjänar som inspiration i denna avhandling för att identifiera nya mekanismer för manipulering av strömning. Samverkan mellan fluider och strukturer (så kallad fluid-struktur-interaktion) kan delas upp i två kategorier. Den första typen av samverkan är aktiv, vilket innebär att en organism aktivt rör hela eller delar av sin kropp för att manipulera det omgivande strömningsfältet (till exempel fåglar som flaxar sina vingar). Den andra typen är passiv samverkan, där organismer har utväxter (svansar, fjärdar, etc.) eller ytbeläggningar som de inte aktivt har kontroll över och som således inte förbrukar någon energi. Ett exempel är fjädrar som passivt rör sig i det omgivande flödet. Vårt mål är att hitta nya passiva mekanismer som växelverkar med den omgivande fluiden på ett fördelaktigt sätt, exempelvis genom att öka lyftkraften eller minska luftmotståndet. I den första delen av detta arbete undersöker vi en enkel modell för en utväxt (i form av en platta) bakom en cirkulär cylinder eller sfär. Om plattan är tillräckligt kort och om det finns ett vak bakom kroppen kommer det upprätta läget av plattan att vara instabilt. Denna instabilitet är i princip samma som uppstår då man försöker balansera en penna på fingret. Vi förklarar den bakomliggande mekanismen av denna instabilitet genom numeriska beräkningar, experiment och en enkel modell med tre frihetsgrader. Konsekvenserna av denna instabilitet är en omorientering (rotation) av kroppen och en sidledsförflyttning av kroppen i förhållande till tyngdkraftens riktning. Denna mekanism kan användas djur och frön för att öka deras förmåga att förflytta eller sprida sig i vatten eller luft. I den andra delen av avhandlingen studerar vi modeller av porösa och elastiska material. Vi använder en mångskalig metod för att modellera det poroelastiska materialet som ett kontinuum. Vi härleder randvillkor för både hastighetsfältet och trycket på gränssnittet mellan den fria fluiden och det poroelastiska materialet. Resultaten som erhållits med de härledda randvillkoren valideras sedan genom direkta numeriska simuleringar (DNS) för både två- och tredimensionella fall. Kontinuumsmodellen av materialet kopplad genom randvillkoren till den fria strömmande fluiden predikterar strömnings- och förskjutningsfält noggrant i jämförelse med DNS.

Much of the vibration in fast moving vehicles is caused bydistributed random excitation, such as turbulent flow and roadroughness. Piping systems transporting fast flowing fluid isanother example, where distributed random excitation will causeunwanted vibration. In order to reduce these vibrations andalso the noise they cause, it is important to have accurate andcomputationally efficient prediction methods available. The aim of this thesis is to present such a method. Thefirst step towards this end was to extend an existing spectralfinite element method (SFEM) to handle excitation of planetravelling pressure waves. Once the elementary response tothese waves is known, the response to arbitrary homogeneousrandom excitation can be found. One example of random excitation is turbulent boundary layer(TBL) excitation. From measurements a new modified Chase modelwas developed that allowed for a satisfactory prediction ofboth the measured wall pressure field and the vibrationresponse of a turbulence excited plate. In order to model morecomplicated structures, a new spectral super element method(SSEM) was formulated. It is based on a waveguide formulation,handles all kinds of boundaries and its elements are easily putinto an assembly with conventional finite elements. Finally, the work to model fluid-structure interaction withanother wave based method is presented. Similar to the previousmethods it seems to be computationally more efficient thanconventional finite elements. / <p>NR 20140805</p>

La rupture de la plaque d'athérome carotidienne est la première cause des infarctus cérébraux. Pour prévenir ces accidents, l'endartérectomie carotidienne est le traitement le plus utilisé. La vulnérabilité de la plaque est en relation avec les efforts que le sang applique sur la plaque. Ces actions sont différentes suivant les propriétés constitutives, mécaniques et géométriques de la plaque. Plusieurs auteurs ont développé des modèles numériques de la plaque d'athérome carotidienne à partir desquels une analyse mécanique a permis de caractériser les déformations et les contraintes en lien avec la rupture de la plaque. Néanmoins, les caractéristiques d'une plaque vulnérable sont encore mal connues. Dans ce manuscrit, un modèle numérique de plaque d'athérome carotidienne est développé en interaction fluide-structure dans le but mieux comprendre comment les actions mécaniques du sang sur la plaque sont affectées par les propriétés mécaniques et géométriques de la plaque. Plusieurs résultats sont en concordance avec la littérature : la vulnérabilité de la plaque est associée à la sévérité de sténose et à l'épaisseur de la chape fibreuse. De plus une analyse de l'écoulement du sang, de la déformation de la plaque et des contraintes dans la plaque révèle que les effets de l'écoulement du sang sont amplifiés si la plaque est courte, si la pente en amont de sténose est raide ou si la morphologie de la plaque est irrégulière et asymétrique. Ces résultats offrent de nouvelles perspectives dans la compréhension de la vulnérabilité de la plaque. / The rupture of carotid atheromatous plaques is the major cause of cerebrovascular thromboembolic events such as strokes and ischemic attacks. To prevent this issue, carotid endarterectomy is the preferred treatment. The vulnerability of the plaque is related the mechanical action of the blood onto the plaque. This action is different according to the plaque morphology, the plaque constitution and the mechanical properties of the constituents. Several authors developed computational models to perform mechanical analyses for carotid atherosclerotic plaques and to identify critical mechanical descriptors as stresses or strains related to plaque rupture. But the question of which plaque characteristics affect the plaque rupture is not closely elucidated. In this manuscript a fluid structure interaction model is developed, questioning how the mechanical action of the blood onto an atheromatous plaque is affected by the mechanical and geometrical properties of the plaque. Many results are in agreement with the literature: the vulnerability of atheromatous plaques is related to the degree of severity of the endoluminal stenosis and the thickness of the fibrous cap. Moreover the resulting flow patterns, wall shear stresses, plaque deformations and stresses in the fibrous cap reveal that the effects of the blood flow are amplified if the plaque is short, if the slope upstream stenosis is steep or if the plaque morphology is irregular and asymmetric. These results offer new perspectives for understanding the vulnerability of plaques.

Cette thèse s’intéresse à la faisabilité de la mise en place de modèles d’ordre réduit pour l’étude des vibrations sous écoulement au sein de faisceaux tubulaires d’échangeurs de chaleur. Ces problématiques sont cruciales car les systèmes étudiés sont des éléments majeurs des centrales nucléaires civiles et des chaufferies embarquées dans les sous-marins.Après avoir rappelé le fonctionnement et les risques vibratoires existants au sein des échangeurs de chaleur, des calculs complets d’écoulement et de vibrations sous écoulement ont été effectués, d’abord pour un tube seul en milieu infini, puis pour un faisceau de tubes. Ces calculs ont été menés avec l’outil CFD Code_Saturne. La méthode de réduction de modèle POD (Proper Orthogonal De-composition) a été appliquée au cas des écoulements avec la structure laissée fixe.Les résultats obtenus montrent l’efficacité de la méthode pour ces configurations, moyennant l’introduction de méthodes de stabilisation pour l’écoulement au sein du faisceau. La méthode POD-multiphasique, permettant d’adapter la méthode POD à l’interaction fluide-structure, a ensuite été appliquée. Les grands déplacements d’un cylindre seul dans la zone d’accrochage (lock-in) ont été correctement reproduits par cette méthode de réduction de modèle. De même, on montre que les grands déplacements d’un cylindre en milieu confiné dans un faisceau de tubes sont fidèlement reconstruits.Enfin, l’extension de l’utilisation de la réduction de modèle aux études d’évolution paramétrique a été testée. Nous avons d’abord utilisé la technique considérant une base POD unique pour reproduire des écoulements à divers nombres de Reynolds autour d’un cylindre seul. Les résultats confirment la prédictivité bornée à une gamme de paramètres de cette méthode. Enfin, l’interpolation de bases POD pré-calculées pour une famille de paramètres donnés, utilisant les variétés de Grassmann et permettant de générer de nouvelles bases POD, a été testée sur des cas modèles. / The objective of this thesis is to study the ability of model reduction for investigations of flow-induced vibrations in heat exchangers tube bundle systems.These mechanisms are a cause of major concern because heat exchangers are key elements of nuclear power plants and on-board stoke-holds.In a first part, we give a recall on heat exchangers functioning and on vi-bratory problems to which they are prone. Then, complete calculations leaded with the CFD numerical code Code_Saturne are carried out, first for the flow around a single circular cylinder (fixed then elastically mounted) and then for the case of a tube bundle system submitted to cross-flow. Reduced-order method POD is ap-plied to the flow resolution with fixed structures. The obtained results show the efficiency of this technique for such configurations, using stabilization methods for the dynamical system resolution in the tube-bundle case.Multiphase-POD, which is a method enabling the adaptation of POD to fluid-structure interactions, is applied. Large displacements of a single cylinder elastically mounted under cross-flow, corresponding to the lock-in phenomenon,are well reproduced with this reduction technique. In the same way, large displace-ments of a confined moving tube in a bundle are shown to be faithfully recon-structed.Finally, the use of model reduction is extended to parametric studies. First,we propose to use the method which consists in projecting Navier-Stokes equations for several values of the Reynolds number on to a unique POD basis. The resultsobtained confirm the fact that POD predictability is limited to a range of parameter values. Then, a basis interpolation method, constructed using Grassmann mani-folds and allowing the construction of a POD basis from other pre-calculated basis,is applied to basic cases.