Le cancer colorectal est marqué par une importante mortalité dans les stades avancés du fait du fort taux de récidives tumorales après chimiothérapie. La prédiction de l'efficacité et de la toxicité des cytotoxiques s'impose comme un des enjeux majeurs de ces prochaines années. Parce que la majorité des anticancéreux sont pris en charge par les enzymes et transporteurs dont l'expression est contrôlée par le niveau d'expression et d'activation des xénosenseurs CAR et PXR, il est fort probable que ces xénosenseurs puissent représenter des facteurs prédictifs à prendre en compte dans la prise en charge des cancers. Notre équipe a récemment montré que les récepteurs des xénobiotiques PXR (NR1I2) (Raynal et al, 2010) et CAR (NR1I3) sont exprimés dans des lignées cellulaires et des tissus coliques humains. Leur surexpression dans les lignées coliques LS174T et T84 entraine leur résistance à l'irinotécan et à son métabolite actif le SN38 alors que leur inhibition antagonise cette résistance. Des dosages intra- et extra-cellulaires du SN38 et du SN38-G, ainsi que la quantification des ARNm des l'UGT1As et du transporteur MDR1, montrent que CAR et PXR augmentent le métabolisme détoxifiant et l'efflux du SN38. L'impact de la surexpression de ces xénosenseurs sur la viabilité des cellules LS174T à différentes classes de cytotoxiques (anti-métabolites, intercalants, inhibiteurs de topoisomérases, poisons du fuseau) a ensuite été évaluée. Nous avons observé que l'expression de CAR ou PXR conduit à une forte chimiorésistance au paclitaxel, au docétaxel et au 4-hydroxy-cyclophosphamide alors que PXR entraîne une sensibilisation marquée au cisplatine et au carboplatine en augmentant la quantité d'adduits de platine sur l'ADN. Les études du transcriptome de nos modèles cellulaires nous ont permis d'identifier les gènes cibles impliqués dans ces variations de cytotoxicité. Des études de confirmation par modulation pharmacologique ou ARNs interférents de ces gènes cibles sont en cours et nous permettront de préciser les mécanismes mis en jeu dans les variations de chimiosensibilité. Ces travaux devraient permettent de mieux appréhender le rôle des xénosenseurs CAR et PXR sur le métabolisme intra-tumoral des cytotoxiques et potentiellement sur la réponse à des chimiothérapies variées. / Colorectal cancer is characterized by high mortality in advanced stages due to the high rate of tumor recurrence after chemotherapy. The prediction of the efficacy and toxicity of cytotoxic drugs represents a major challenge in the coming years. Because the majority of cancer drugs are supported by the enzymes and transporters whose expression is controlled by the level of expression and activation of the xenosensors CAR (NR1I3) and PXR (NR1I2), it is likely that they may represent predictive factors in the management of cancer. Our team has recently shown that xenobiotic receptors PXR (Raynal et al, 2010) and CAR are expressed in cell lines and human colon tissues. Their overexpression in colon cancer cell lines LS174T and T84 leads to resistance to irinotecan and to its active metabolite SN38, while their inhibition reverse this resistance. Irinotecan metabolites detection assays of SN38 and SN38G, and the quantification of UGT1As and MDR1 mRNA, show that CAR and PXR increase the detoxifying metabolism and the efflux of SN38. The impact of overexpression of these xenosensors on LS174T cell viability to different classes of cytotoxic agents (anti-metabolites, DNA intercalators, topoisomerase inhibitors, antimitotic agents) was then evaluated. We observed that the expression of CAR or PXR results in a significant drug resistance to paclitaxel, docetaxel and 4-hydroxy-cyclophosphamide whereas PXR leads to a marked sensitization to cisplatin and carboplatin by increasing the amount of platinum adducts the DNA. Microarray studies of our cell models allowed us to identify the target genes potentially involved in these changes in cytotoxicity. Further studies by pharmacological modulation or interfering RNAs of these target genes are in progress and will allow us to clarify the mechanisms involved in the changes in chemosensitivity. This work should help us to understand the impact of CAR and PXR xenosensors on the intratumoral metabolism of cytotoxic drugs and potentially on the response to various chemotherapies.

La conduite alimentaire n'obéit pas nécessairement au besoin physiologique de consommer des aliments (la faim) ou à la satiété suggérant qu'un système nerveux volontaire de la restriction (anorexie) et de la consommation excessive d'aliments (boulimie, « binge-type eating ») inhibe le système nerveux autonome. Ces deux anomalies affectent plus fréquemment, et souvent à la fois, la femme que l'homme. Si l'utilisation de modèles animaux permet l'étude d'une part des bases neuronales en cause, ceux possiblement responsables de l'alternance « anorexie / boulimie » restent encore à identifier. Dans ce contexte, nous avons ainsi centré nos analyses sur l'étude de l'implication de récepteurs cérébraux couplés aux protéines G ; les récepteurs 4 de la sérotonine (R5-HT4) car leur stimulation dans le noyau accumbens (NAc), une structure du système de la récompense, inhibe la faim y compris après la mise à jeun de souris, par l'action AMPc/PKA dépendante d'un peptide de l'addiction ; « cocaine- and amphetamine-regulated transcript » (CART). Nous montrons que le maintien d'une plus forte expression (ectopique ou physiologique) des R5-HT4 dans le NAc a réduit plus durablement la faim que sa seule stimulation et augmente l'activité locomotrice. En incluant dans notre raisonnement l'activité constitutive des R5-HT4 (e.g. accumulation de la forme active R*), nous montrons que l'injection d'un agoniste inverse (inhibition de l'activité constitutive : accumulation de la forme inactive, R) spécifique des R5-HT4 dans le NAc entraîne une baisse du taux d'AMPc et de CART tout en augmentant celui des ARNm codant le NPY d'autant plus que l'hyperphagie est élevée. Les effets induits par l'injection de l'agoniste inverse ne sont pas observés lorsqu'il est adjoint à un antagoniste des R5-HT4. Ces résultats suggèrent une implication physiologique de l'activité constitutive des R5-HT4 dans la régulation de la conduite alimentaire; son inhibition (agoniste inverse) dans le NAc augmente la prise et reprise alimentaire après un jeûne. L'ensemble de ces résultats rend probable que la plus forte activité des R5-HT4, à la base d'une association « anorexie /hyperactivité locomotrice », souvent décrite comme paradoxale au plan énergétique dans le syndrome de l'anorexie mentale, représente plutôt un mécanisme de compensation globale d'une valeur énergétique à perdre en conséquence d'une trop forte consommation d'aliments. Puisque la densité des R5-HT4 peut varier selon un taux variable de 5-HT après stress, lequel aggrave les anomalies alimentaires, nous avons étudié plus avant l'implication des R5-HT4 dans l'effet anorexigène du stress (immobilisation forcée) chez des souris femelles privées de leur gène : l'hypophagie induite par le stress n'a pas été observée chez les souris privées des R5-HT4 qui présentent une possible hyperactivité de l'axe hypothalamo-hypophysaire corticosurrénalien vraisemblablement compensée par un plus fort rétrocontrôle négatif. Il est donc probable que les R5-HT4 contribuent à réduire les conséquences du stress et que la modification de l'équilibre de leur activité contribue à une part de la symptomatologie de patients atteints d'anorexie / boulimie. / Feeding behavior does not necessarily obey to the physiological need to eat (hunger) or to satiety, suggesting that voluntary nervous system of the restriction (anorexia) and overeating (bulimia, binge-type eating) inhibits the autonomic nervous system. These two anomalies affecting more frequently, and often both, the woman than man. If animal models are used to study a part of neural bases involved, those possibly responsible for the oscillation of"anorexia / bulimia" remain to be identified. In this context, we thus focused our analysis on the study of the involvement of brain receptors coupled to G proteins ; serotonin 4 receptors (5-HTR4) because their stimulation in the nucleus accumbens (NAc), a brain reward area, inhibits hunger even after a food deprivation, by the action ofcAMP / PKA, dependent of an addiction peptide, "cocaine-and amphetamine-regulated transcript" (CART). Weshow that maintaining a higher expression (ectopic or physiological) of 5-HTR4 in the NAc, reduced hunger more longer than the acute stimulation and increased locomotor activity. Including in our reasoning the constitutive activity of 5-HTR4 (e.i. accumulation of the active form R*), we show that injecting a specific inverse agonist of the5-HTR4 (inhibition of constitutive activity: accumulation of inactive form, R ) in the NAc induced a decrease incAMP and CART levels, while increasing NPY mRNA level, especially when binge is high. The effects induced by the injection of the inverse agonist are not observed when a 5-HTR4 antagonist was coadministrated. These results suggest a physiological involvement of the constitutive activity of 5-HTR4 in the regulation of feeding behavior ; its inhibition (inverse agonist) in the NAc increases the food intake in fed or food-deprived mice. All of these results makes it likely that the highest activity of 5-HTR4, at the base of the association "anorexia /locomotor hyperactivity", often described as paradoxical in terms of energy, in the syndrome of anorexia nervosa,represent rather a global compensation mechanism of energy to be lost as a result of an excessive consumption of food. Since the density of the 5-HTR4 may vary depending on a variable rate of 5-HT following stress, which aggravates the feeding disorders, we further investigated the involvement of 5-HTR4 in the appetite-suppressant effect of stress (forced immobilization) in female mice deprived of their gene: stress-induced hypophagia was not observed in mice deprived of 5-HTR4 who present a possible hyperactivity of the hypothalamic-pituitary adrenocortical axis likely offset by a stronger negative feedback. It is therefore likely that the 5-HTR4 contribute to reduce the effects of stress and that the modification of the balance of their activities contribute to a part of the symptoms of patients with anorexia / bulimia.

Les noyaux des cellules eucaryotes contiennent différents compartiments définis fonctionnellement ou structurellement et à multiples échelles qui présentent une distribution spatiale très ordonnée. Un défi majeur est alors d'identifier les règles selon lesquelles les compartiments nucléaires sont organisés dans l'espace et de décrire comment ces règles peuvent varier en fonction des conditions physiologiques ou expérimentales. Les statistiques spatiales ont rarement été utilisées à cette fin et se sont généralement limitées à l'évaluation de l'aléatoire complet. Ici, nous développons une approche de statistiques spatiales qui combine la cytologie, l'analyse d'image et la modélisation spatiale pour mieux comprendre les configurations spatiales à l'intérieur du noyau. Notre première contribution est un cadre méthodologique qui permet de tester des modèles d'organisation spatiale au niveau de la population. Plusieurs modèles spatiaux ont été proposés et mis en œuvre, en particulier l'aléatoire, l'aléatoire orbitale, la régularité maximale, l'aléatoire territoriale et l'aléatoire territoriale orbitale, pour analyser la distribution d'objets biologiques dans des domaines 3D finis et de formes arbitraires. De nouveaux descripteurs spatiaux, combinés aux descripteurs classiques, sont utilisés pour comparer les motifs observés à des configurations attendues sous les modèles testés. Une version non biaisée d'un test statistique publié précédemment est proposé pour évaluer la qualité de l'ajustement des modèles spatiaux sur les distributions observées. Après, nous étudions les propriétés de l'approche proposée à partir de données réelles et simulées. La robustesse de l'approche proposée aux erreurs de segmentation et la fiabilité des évaluations spatiales sont examinées. En outre, la base d'une méthode pour comparer les distributions spatiales entre différents groupes expérimentaux est proposée. Dans la dernière partie de ce travail, notre approche est appliquée à des images de noyaux cellulaires de la feuille chez A. thaliana, pour analyser la distribution spatiale de compartiments dynamiques et plastiques d'hétérochromatine, les chromocentres, qui ont un rôle important dans la structure du génome. Des noyaux isolés et des cryo-sections provenant de plantes de type sauvage ont été analysés. Nous montrons que les chromocentres présentent une distribution très régulière, ce qui confirme les résultats publiés précédemment. En utilisant nos nouveaux descripteurs, nous démontrons pour la première fois, objectivement et quantitativement, que les chromocentres présentent une localisation préférentielle périphérique. En employant un nouveau modèle spatial, nous rejetons l'hypothèse selon laquelle l'organisation régulière observée est uniquement expliquée par un positionnement périphérique. Nous démontrons en outre que les chromocentres affichent une régularité spatiale proche de la régularité maximale à l'échelle globale, mais pas à l'échelle locale. Enfin, nous utilisons des simulations pour tester un modèle selon lequel le positionnement des chromocentres est déterminé par les territoires chromosomiques et par des interactions avec l'enveloppe nucléaire. Nous avons ensuite vérifié s'il la distribution des chromocentres pouvait être modifiée par différentes mutations. Nous avons analysé les données de noyaux des mutants crwn et kaku, qui sont connus pour influencer la morphologie nucléaire. Les résultats suggèrent que ces mutations impactent en effet la morphologie nucléaire et les caractéristiques de l'hétérochromatine, mais ne modifient pas la régularité de la distribution des chromocentres même si la distance à la frontière du noyau est affectée. La généricité du cadre proposé pour analyser les distributions d'objets dans des domaines 3D finis et la possibilité d'ajouter de nouveaux modèles et descripteurs spatiaux devrait permettre d'analyser des organisations spatiales au sein de différents systèmes biologiques et à différentes échelles. / Eukaryotic cell nuclei contain distinct functionally or structurally defined compartments at multiple scale that present a highly ordered spatial arrangement. Several studies in plants and animals have pointed out to links between nuclear organization and genome functions. A key challenge is to identify rules according to which nuclear compartments are organized in space and to describe how these rules may vary according to physiological or experimental conditions. Spatial statistics have been rarely used for this purpose, and were generally limited to the evaluation of complete spatial randomness. In this Thesis, we develop a spatial statistical approach, which combines cytology, image analysis and spatial modeling to better understand spatial configurations inside the nucleus. Our first contribution is a methodological framework that allows to test models of spatial organization at the population level. Several spatial models have been developed and implemented, including randomness, orbital randomness, maximum regularity, territorial randomness or orbital territorial randomness of biological objects within finite 3D domains of arbitrary shape. New spatial descriptors, in combination with classical ones, are used to compare observed patterns to expected configurations under the tested models. An unbiased version of a previously published statistical test is proposed to evaluate the goodness-of-fit of spatial models over populations of observed patterns. In the second part of this Thesis, we investigate the properties of the proposed approach using simulated and real data. The robustness of the proposed approach to segmentation errors and the reliability of the spatial evaluations are examined. Besides, the basis for a method to compare spatial distributions between different experimental groups is proposed. In the last part of this work, the proposed approach is applied on A. thaliana leaf cell nuclei images to analyze the spatial distribution of chromocenters, which are dynamic and plastic heterochromatic compartments that have an important structural role in the genome. A first application was to analyze isolated and cryo-section nuclei from wild type plants. We show that chromocenters present a highly regular distribution, confirming previously published results. Using new spatial statistical descriptors, we then demonstrate objectively and quantitatively, for the first time, that chromocenters exhibit a preferentially peripheral localization. Employing a new spatial model, we then reject the hypothesis that the regular organization is explained solely by the peripheral positioning. We further demonstrate that chromocenters organization displays a close-to-maximum spatial regularity at the global scale, but not at the local one. Lastly, we use simulations to examine a model according to which chromocenters positioning is constrained by chromosome territories and by interactions with the nuclear boundary. The second application was to elucidate whether chromocenters distribution could be altered under different mutations. We analyze nuclei data from crwn and kaku mutants, which are known to affect nuclear morphology. The results suggest that these mutations impact on nuclear morphology and on heterochromatin features but do not alter the regularity of chromocenters distribution even when the relative distance to the border is affected. The genericity of the proposed framework to analyze object distributions in finite 3D domains and its expandability to add more spatial models and spatial descriptors should be of interest to decipher spatial organizations within biological systems at various scales.

Geosynthetic reinforced soil (GRS) is a soil improvement technology in which closely spaced horizontal layers of geosynthetic are embedded in a soil mass to provide lateral support and increase strength. GRS is popular due to a relatively new application for bridge support, as well as long-standing application in mechanically stabilized earth walls. Several different GRS design methods have been used, and some are application-specific and not based on fundamental principles of mechanics. Because consensus regarding fundamental behavior of GRS is lacking, numerical and mathematical analyses were performed for laboratory tests obtained from the published literature of GRS under triaxial compression in consolidated-drained conditions. A three-dimensional numerical model was developed using FLAC3D. An existing constitutive model for the soil component was modified to incorporate confining pressure dependency of friction angle and dilation parameters, while retaining the constitutive model's ability to represent nonlinear stress-strain response and plastic yield. Procedures to obtain the parameter values from drained triaxial compression tests on soil specimens were developed. A method to estimate the parameter values from particle size distribution and relative compaction was also developed. The geosynthetic reinforcement was represented by two-dimensional orthotropic elements with soil-geosynthetic interfaces on each side. Comparisons between the numerical analyses and laboratory tests exhibited good agreement for strains from zero to 3% for tests with 1 to 3 layers of reinforcement. As failure is approached at larger strains, agreement was good for specimens that had 1 or 2 layers of reinforcement and soil friction angle less than 40 degrees. For other conditions, the numerical model experienced convergence problems that could not be overcome by mesh refinement or reducing the applied loading rate; however, it appears that, if convergence problems can be solved, the numerical model may provide a mechanics-based representation of GRS behavior, at least for triaxial test conditions. Three mathematical theories of GRS failure available in published literature were applied to the laboratory triaxial tests. Comparisons between the theories and the tests results demonstrated that all three theories have important limitations. These numerical and mathematical evaluations of laboratory GRS tests provided a basis for recommending further research. / Ph. D.

The preservative challenge test is a regulatory requirement specified in various pharmacopoeias to determine the efficacy of preservatives. However, such testing is a labour-intensive repetitive task and often requires days before results can be generated. Microbial biosensors have the potential to provide a rapid and automated alternative to the traditional viable counting currently in use. However, the selection of appropriate promoters is essential. The bioluminescent reporter strains used in the current study comprise the Photorhabdus luminescence lux CDABE reporter genes under the control of five individual constitutive Escherichia coli promoters: outer lipoprotein (lpp); twin arginine translocase (tatA); lysine decarboxylase (ldc); lysyl t-RNA (lysS); and ribosomal protein (spc). The promoter plus lux CDABE constructs were cloned, ligated into the plasmid vector pBR322 and transformed into E. coli ATCC 8739. The bioluminescence intensity in the decreasing order of constitutive promoter was lpp > spc> tatA> ldc > lysS. The five biosensor strains tested successfully in PET assays and demonstrated accuracy with a minimum detection limit of 103 CFU/ml, a detection range of 6 orders magnitude, and yielded equivalent results to methods currently recommended by the pharmacopoeias. The bioluminescent biosensors were used to monitor the efficacy of preservatives; sorbic acid at concentrations of 0.031% to 0.2% at pH 5.0, and benzalkonium chloride at concentrations of 0.0062% to 0.00039% alone and in combination with 0.03% EDTA. The 99.9% percentage of bioluminescence reduction of tatA-lux, ldc-lux, lysS-lux, and spc-lux was statistically equivalent to the 3 log10 CFU/ml reduction as required by the Pharmacopeias’. Strong significant correlations between bioluminescence and the methods recommended by the pharmacopoeias were obtained when the biosensor strains were challenged with preservatives, for all except lpp-lux E. coli. The bioluminescence expressed by the lpp-lux biosensor was significantly lower during long-term stationary phase than it was for any of the other biosensors and was also significantly lower than for any of the other biosensors in the presence of preservatives. Since the plasmid copy number and viable counts for lpp-lux did not change under these conditions, it suggests that perhaps lpp-lux was down regulated under stress conditions. There were no statistically significant differences between the results of the bioluminescence assays and the results of the viable count and ATP chemiluminescence assay. Virtual foot printing (using Regulon DB database) demonstrated that two crp binding sites overlapping the -10 regions are located on the negative strand of the lysS promoter sequences and that one crp binding site is located in lpp. The biosensor strains ldc-lux exhibited levels of bioluminescence per cell significantly lower than spc in the presence of preservatives whilst there was a significant increase in bioluminescence per cell by tatA-lux under alkaline conditions (pH 8.9) during long-term stationary phase. Amongst the five biosensor strains tested in the current work, it was determined that the spc-lux strain would be the most attractive candidate for further work, since the bioluminescence expressed per cell was significantly greater, by 10-1000 times, than that expressed by the other four promoters when challenged with the preservatives tested with excellent significant correlations between bioluminescence expression and viable counts in the PET assays with the various preservatives in this study (R2: 8.79-1.00). The bioluminescent biosensor strains showed no statistical differences from the control strains (wildtype E.coli ATCC 8739 and E.coli carrying a promoterless [pBR322.lux] for adneylate energy charge (AEC), plasmid copy number (PCN) bioluminescence or viable counts over 28 days. The emission of bioluminescence by the four bioreporter strains across 28 days is reflected by the stability of PCN with correlations of 0.78-0.90, except for lpp-lux with R2: 0.59. The following promoter elements were found likely to assist greater expression of bioluminescence: an A+T level of approximately 50% between the -40 and -60 regions (the UP element); a G+C level of approximately 50% within the -10 and +1 regions; the extended -10 region and -10 region of consensus sequence RpoD (σ70/D).

This work presents an effective methodology for measuring the depth-resolved 3D full-field deformation of semitransparent, light scattering soft tissues such as vertebrate eye cornea. This was obtained by performing digital volume correlation on optical coherence tomography volume reconstructions of silicone rubber phantoms and porcine cornea samples. Both the strip tensile tests and the posterior inflation tests have been studied. Prior to these tests, noise effect and strain induced speckle decorrelation were first studied using experimental and simulation methods. The interpolation bias in the strain results has also been analyzed. Two effective approaches have been introduced to reduce the interpolation bias. To extract material constitutive parameters from the 3D full-field deformation measurements, the virtual fields method has been extended into 3D. Both manually defined virtual fields and the optimized piecewise virtual fields have been developed and compared with each other. Efforts have also been made in developing a method to correct the refraction induced distortions in the optical coherence tomography reconstructions. Tilt tests of different silicone rubber phantoms have been implemented to evaluate the performance of the refraction correction method in correcting the distorted reconstructions.

This dissertation work focuses on the thermomechanical behaviors of two recent exciting developments in active polymers: shape memory (SM) effects and covalent adaptive network polymers with bond exchange reactions. Both polymers are active in performing prescribed functions when an external stimulus is applied. The goals of the studies are to understand complex thermomechanical behaviors of such smart polymers through experiments, develop constitutive models to describe the behaviors, and use the developed models to assist their development and engineering applications. For the polymer SM effect, we use a multi-branched constitutive model to study the SM effect achieved by polymer glass transition. The major finding of our study is that the “Reduced Time” is identified to be the unique parameter to determine the polymer shape fixity and recovery ratio under different thermo-temporal conditions in an SM cycle. Based on the theoretical knowledge, we also study the energy releasing mechanism within shape memory polymers (SMPs), multi-shape memory effects, as well as the SM properties in various composite systems, such as magnetic particles, carbon black and microvascular reinforced SMP composites. For the covalent adaptive network polymers, we adopt the emerging covalent chemistry BERs to achieve a malleable, reparable, recyclable and yet insoluble thermoset network. After being pulverized into micro-size, and then compressed either at high temperature or just facilitated by the moisture, the polymer powder could be welded on the interfaces, and assembled together into a new sample with comparable mechanical properties to the fresh sample. Theoretical models are developed to gain fundamental understanding of how the processing conditions can affect the quality of reprocessed materials. A molecular model is developed to understand welding kinetics at the interface. Such understanding is then used to develop a multiple length scale interfacial constitutive model, which can be implemented in to finite element simulation software to assist computational study of reprocessing process.

The main weakness of soil is its inability to resist tensile stresses. Civil engineers have been trying to address this problem for decades. To increase the tensile and shear strengths of soil, different methods of reinforcing such as using geosynthetics have been used in different types of earth structures such as retaining walls, earth dams, slopes, etc. Due to the excellent corrosion resistance of polymers, the use of geosynthetics has increased dramatically in recent years. However, there are some significant problems associated with geosynthetics, such as creep and low modulus of elasticity. In this research, a new Geo-Composite which is made of Carbon Fiber Reinforced Polymer (CFRP) is used to overcome some of the short comings of the existing geosynthetics. The new Geo-Composite has all the benefits of the geotextiles plus higher strength, higher modulus and no creep. In first part of the investigation, over eighty experiments were carried out using direct shear test. The interface properties of the Geo-Composite (CFRP) and fine sand were investigated. Tests showed that the interface shear behavior between Geo-Composite and fine sand depended on the normal forces during the curing of epoxy and curing age of epoxy. The two methods used to prepare the specimen are pre-casting and casting in place, and the results of these two methods are compared. In the second part of the investigation, the pull-out test device was designed and assembled using a triaxial loading device and a direct shear device. In the pull-out test, the normal force applied by the triaxial loading and pull out force is applied by a direct shear device. CFRP samples were prepared in the lab, and pre-cast and cast-in-place samples were tested using fine sand. The pull-out force and corresponding displacements of each of the materials were recorded and compared. In the third part of the investigation, the behavior of the interface between coarse sand and modified CFRP has been studied in larger scale using a device known as Cyclic Multi Degree of Freedom (CYMDOF) device. A constitutive Model, Hierachical Single Surface (HISS) model, is used to characterize the behavior of the interfaces. The constitutive model is verified by predicting the laboratory behavior of interface. In the forth part of the investigation, using the laboratory test data results, a finite element procedure with the hardening model is used to simulate field behavior of a CFRP reinforced earth retaining wall, and compare the results with a geotextile reinforced earth retaining wall. This section shows the advantages and disadvantages of using CFRP in MSE walls.

Plastic packaging waste currently forms a significant part of municipal solid waste and as such is causing increasing environmental concerns. Such packaging is largely non-biodegradable and is particularly difficult to recycle or to reuse due largely to its complex compositions. Apart from limited recycling of some easily identifiable packaging wastes that can be separated economically, such as bottles, most packaging waste ends up in landfill sites. In recent years, in an attempt to address this problem in plastic packaging, the development of packaging materials from renewable plant resources has received increasing attention and a wide range of bioplastic materials based on starch are now available. Environmentally these bioplastic materials also reduce reliance on oil resources and have the advantage that they are biodegradable and can be composted upon disposal to reduce the environmental impact. Many food packaging containers are produced by thermoforming processes in which thin sheets are inflated under pressure into moulds to produce the required thin -wall structures. Hitherto these thin sheets have almost exclusively been made of oilbased polymers and it is for these that computational models of thermoforming processes have been developed. Recently, in the context of bioplastics, commercial thermoplastic starch sheet materials have been developed. The behaviour of such materials is influenced both by temperature and, because of the inherent hydrophilic characteristics of the materials, by moisture content. Both of these aspects affect the behaviour of bioplastic sheets during the thermoforming process. This thesis describes experimental work and work on the computational modelling of thermoforming processes for thermoplastic starch sheets using a commercially available material. The experimental work has been carried in order to characterise the deformation behaviour of the material with regard to different temperature, moisture contents and strain rates. Thermoforming of the material was performed and samples produced were used for comparison and verification of the computational modelling of the thermoforming process. In the first attempt to model the thermoforming process, a hyperelastic constitutive equation was established to approximate the material behaviour taking account of the combined effects of temperature and moisture content and a simple ii membrane model with constrained deformation was used to model an axisymmetric case of thermoforming. Simulations with this model showed that moisture content mostly affects the pressure required to push the sheet into the mould while moisture variation during thermoforming has little effect on the final thickness distribution of the product. Considerable discrepancies were found in the thickness distribution between the predictions from the model and the experimental measurements. Further attempts were made to take account of the elasto-plastic behaviour of the material and a more complex three-dimensional FE model was developed using ANSYS/LS-DYNA. Based on the findings in the simpler modelling work, no attempt was made to incorporate the moisture content effect on material behaviour but the material parameters for the elasto-plastic constitutive equation were obtained from high speed tensile tests so that moisture variation during thermoforming could be minimised and neglected. The predictions from this model have led to significant improvements in prediction of the thickness distribution which has become much closer to the experimental measurements in comparison with the hyperelastic model. This work provides some important insights into thermoforming of thermoplastic starch materials: a) Deformation behaviour of such materials depends strongly on the moisture content and the temperature, both of which affect behaviour during thermoforming processes, including the preheating stage; b) moisture variation during the thermoforming process has a significant effect on the pressure required for the deformation. This also leads to variation of moisture content distribution in the final product, which in turn affects the material properties such as ductility or impact strength at different positions in the thermoformed structure; c) thermoforming of thermoplastic starch materials can be simulated more accurately by an elasto-plastic model and the LS-DYNA algorithm in comparison with a hyperelastic membrane model. This work has provided useful information on thermoforming of thermoplastic starch materials with particular reference to the design of thermoforming tools and to the careful control of processing conditions including preheating. It has also laid a solid foundation for future work on how the moisture variation impacts on the formation of defects such as incomplete forming due to material hardening and fracture due to loss of ductility.

Over the past decades simulation techniques, and in particular finite element method, have been used successfully to predict the response of systems across a whole range of industries including aerospace, automotive, chemical processes, geotechnical engineering and many others. In these numerical analyses, the behaviour of the actual material is approximated with that of an idealised material that deforms in accordance with some constitutive relationships. Therefore, the choice of an appropriate constitutive model that adequately describes the behaviour of the material plays an important role in the accuracy and reliability of the numerical predictions. During the past decades several constitutive models have been developed for various materials. In recent years, by rapid and effective developments in computational software and hardware, alternative computer aided pattern recognition techniques have been introduced to constitutive modelling of materials. The main idea behind pattern recognition systems such as neural network, fuzzy logic or genetic programming is that they learn adaptively from experience and extract various discriminants, each appropriate for its purpose. In this thesis a novel approach is presented and employed to develop constitutive models for materials in general and soils in particular based on evolutionary polynomial regression (EPR). EPR is a hybrid data mining technique that searches for symbolic structures (representing the behaviour of a system) using genetic algorithm and estimates the constant values by the least squares method. Stress-strain data from experiments are employed to train and develop EPR-based material models. The developed models are compared with some of the existing conventional constitutive material models and its advantages are highlighted. It is also shown that the developed EPR-based material models can be incorporated in finite element (FE) analysis. Different examples are used to verify the developed EPR-based FE model. The results of the EPR-FEM are compared with those of a standard FEM where conventional constitutive models are used to model the material behaviour. These results show that EPR-FEM can be successfully employed to analyse different structural and geotechnical engineering problems.