<p>The design and discovery of novel materials are difficult not only due to expensive and time- consuming calculation and measurements of their properties, but also thanks to the infinite search spaces. With the increasingly abundant data from experiments and simulations, learning from data has the potential of bypassing complex physics-based simulations and experiments and providing fast approximations of the solution. Deep learning models are helpful in the design process that requires prohibitively expensive iterative computations. In addition, as efficient and accurate sur- rogate models, trained deep networks can incorporate techniques, such as sensitivity analysis and active learning, to provide guidance in searching promising candidates. Moreover, deep learning models need to account for the material structural information, such as molecule and atom align- ments, chemical bonds, and grain-level interactions, as it plays an important role in determining the macroscopic properties. In this thesis, we start with developing two standard deep learning model- based materials design frameworks for lithium-ion batteries and thermoelectric materials, and we then investigate the feasibility of standard deep learning models on data with graph-structured in- formation and identify the challenges. Finally, we propose a deep graph operator network that effectively capture the spatial dependency encoded in the graph structure to solve networked dy- namical systems.</p> <p><br></p> <p>In the first half of the thesis, we propose a hybrid convolutional neural network to infer lithium- ion battery microstructure properties, Bruggeman’s exponent and shape factor, given its voltage vs. capacity curves. The trained model accurately predicts the microstructural properties on both experimental and simulation data, and it can readily accelerate the processing-properties- performance and degradation characteristics of the existing and emerging chemistries of lithium- ion batteries. Also, we develop a AI-guided framework to discover and design thermoelectric materials, where we train classifiers based on the materials chemical and structural information embeddings and combine with variance-based sensitivity analysis to suggest candidates and con- duct fast screening.</p> <p><br></p> <p>In the second half of the thesis, we build a data-centric framework with a recurrent neural network-based classifier to achieve traffic incident detection on highway networks. We incorporate weak supervised learning and design labeling functions to create large amount of training data with probabilistic labels. The trained deep ensemble accurately detects incidents with predictive uncertainty. To capture the structural information in the network, we then propose a deep graph operator network that maps the input graph state function to the output graph state function. The proposed model enables resolution-independence and zero-shot transfer, where we do not require a set of fixed sensors to encode the graph trajectory and can use the trained model directly on larger graphs with high accuracy. We utilize the proposed model to solve power grid transient stability prediction and traffic forecasting problems.</p>

A numerical simulation model has been developed which describes the manufacture of composite laminates from fiber-reinforced polyimide (PMR-15) matrix prepreg materials. The simulation model is developed in two parts. The first part is the volatile formation model which simulates the production of volatiles and their transport through the composite microstructure during the imidization reaction. The volatile formation model can be used to predict the vapor pressure profile and volatile mass flux. The second part of the simulation model, the consolidation model, can be used to determine the degree of crosslinking, resin melt viscosity, temperature, and the resin pressure inside the composite during the consolidation process. Also, the model is used to predict the total resin flow, thickness change, and total process time. The simulation model was solved by a finite element analysis. Experiments were performed to obtain data for verification of the model. Composite laminates were fabricated from ICI Fiberite HMF2474/66C carbon fabric, PMR-15 prep reg and cured with different cure cycles. The results predicted by the model correlate well with experimental data for the weight loss, thickness, and fiber volume fraction changes of the composite. An optimum processing cycle for the fabrication of PMR-15 polyimide composites was developed by combining the model generated optimal imidization and consolidation cure cycles. The optimal cure cycle was used to manufacture PMR-15 composite laminates and the mechanical and physical properties of the laminates were measured. Results showed that fabrication of PMR-15 composite laminates with the optimal cure cycle results in improved mechanical properties and a significantly reduced the processing time compared with the manufacturer's suggested cure cycle. / Ph. D.

Titanium (Ti) foams of different densities 1622-4100 Kgm-3 made by a powder sintering technique were studied as to their structural and mechanical properties. The foams were tested under static and dynamic loading. The material was tested quasi statically and dynamically under strain rates in the range of 0.001-2500 s-1 and under different loading modes. It was found that strain rate sensitivity is more pronounced in lower density foams. Experiments were complimented by virtual testing. Based on the Voronoi tessellations a computational method was developed to generate stochastic foam geometries. Statistical control was applied to produce geometries with the microstructural characteristics of the tested material. The generated structures were numerically tested under different loading modes and strain rates. Voronoi polyhedrals were used to form the porosity network of the open cell foams. The virtually generated foams replicated the geometrical features of the experimentally tested material. Meshes for finite element simulations were produced. Existing material models were used for the parent material behaviour (sintered Ti) and calibrated to experiments. The virtual foam geometries of different densities were numerically tested quasi statically under uniaxial, biaxial and triaxial loading modes in order to investigate their macroscopic behaviour. Dynamic loading was also applied for compression. Strain rate sensitive and insensitive models were used for the parent material model in order to examine the influence of geometry and material strain rate sensitivity under high rates of deformation. It was found that inertial effects can enhance the strain rate sensitivity for low density foams and numerical predictions for the generated foam geometries were in very good agreement with experimental results. Power laws were established in scaling material properties with density. The study includes: 1. Information on the material behaviour and data for macroscopically modelling this type of foams for a range of densities and under different strain rates. 2. A proposed method for virtually generating foam geometries at a microscopic scale and examine the effect of geometrical characteristics on the macroscopic behaviour of foams.

Les méthodes de sélection de matériaux monolithiques peuvent conduire à des impasses lorsque les exigences fonctionnelles sont très élevées ou contradictoires. Le passage aux multimatériaux peut alors être envisagé. L’objectif de la thèse est de proposer une méthode de conception de multimatériaux à architecture fixée, avec en perspective une application à la conception de conduites offshore pour le génie pétrolier. Seuls les multimatériaux à architecture multicouche sont considérés et la méthode de conception est redéfinie comme une méthode de sélection des constituants du multimatériau et de dimensionnement. Une adaptation des étapes classiques de sélection des matériaux conduit à présenter la méthode en détail sur des exemples simples. Les techniques de programmation par satisfaction de contraintes s’avèrent nécessaires pour la résolution de cas réels de conception multimatériaux. L’application à la conception de conduites offshore permet de valider la méthode et de démontrer sa pertinence. / When the design requirements are either too stringent or are conflicting, no monolithic material solution exists. In such cases the selection of a multimaterial could be considered. The primary aim of this thesis is to provide a methodology for designing multi-materials with a prescribed arrangement of the constituent materials. The second objective is to apply this new methodology to the design of a submarine pipeline. From amongst the huge variety of multi-material arrangements available, this study focusses on multilayered stackings and therefore the design methodology becomes a method for selecting the materials of the stack and sizing the layers. This original approach is presented in detail using basic examples in order to match the steps of classical methods for selecting engineering materials. The constraints programming techniques were very useful for solving real multimaterial design problems. Applying this new method to the design of a submarine pipeline permits its validation and provides proof of its relevance.

The increasing prevalence of digital fabrication technologies and the emergence of a novel materiality in contemporary craft practice have created the need to redefine the critical context of digital jewellery and wearable futures. Previous research in this area, such as that presented by Sarah Kettley (2007a) and Jayne Wallace (2007), has provided the foundations for further enquiry but has not been advanced significantly since its inception. The artistic research presented in this thesis focuses on how smart materials and microelectronic components could be used to create synergetic digital jewellery objects and wearable futures that reflect changes in the body of their wearer and their environment through dynamic responses. Laying the foundations for a theory of <i>Interactive</i> <i>Craft</i> through evaluating different aspects of creative practice that relate to responsive objects with a close relationship to the human body is at the centre of this enquiry. Through identifying four distinct categories of wearable object, the <i>Taxonomy of the Wearable Object</i> is formulated and clearly delineates the current existing conceptual, technological and material perspectives that govern the relationships between different types of wearable objects. A particular focus is placed on exploring the concept of <i>Digital Enchantment</i> and how it could be utilised to progress towards developing the <i>Uncanny Object</i> that appears to possess biological characteristics and apparent agency, yet is a fully artificial construct. The potential for the practical application of a design methodology guided by playful engagement with novel materials, microelectronics and digital fabrication technologies is analysed, taking into account Ingold’s concept of the <i>textility of making </i>(Ingold, 2011). Through exploring the notion of the <i>Polymorphic</i> <i>Practitioner</i> in the context of <i>Alchemical Practice</i>, a model for experiential knowledge generation through engaging in cross-disciplinary collaboration is developed. This is supported by a qualitative survey of European materials libraries, including accounts of site visits that evaluate the usefulness of materials libraries for creative practitioners invested in novel materiality as well as visually documenting a selection of the visited libraries’ most intriguing material holdings. Utilising a scientific testing protocol, a practical body of work that centres on conducting extensive experiments with smart materials is developed, with a particular focus on testing the compatibility and colour outcomes of chromic pigments in silicone. The resulting chromic silicone samples are collated, together with sourced smart materials, in a customised materials library. Investigational prototypes and the <i>Microjewels</i> collection of digital jewellery and wearable futures that responds to external and bodily stimuli whilst engaging the wearer through playful interaction are presented as another outcome of this body of research.

A novel way to ensure stability of mortarless structures – topological interlocking – is examined. In this type of interlocking the overall shape and arrangement of the building blocks are chosen in such a way that the movement of each block is prevented by its neighbours. (The methodological roots of topological interlocking can be found in two ancient structures: the arch and the dry stone wall.) The topological interlocking proper is achieved by two types of blocks: simple convex forms such as the Platonic solids (tetrahedron, cube, octahedron, dodecahedron and icosahedron) that allow plate-like assemblies and specially engineered shapes of the block surfaces that also allow assembling corners. An important example of the latter – so-called Osteomorphic block – is the main object of this research with some insight being provided by numerical modelling of plates assembled from tetrahedra and cubes in the interlocking position. The main structural feature of the interlocking assemblies is the need of the peripheral constraint (for the Osteomorphic blocks this requirement can be relaxed to uni-directional constraint) to keep their integrity. We studied the least visible constraint structure – internal pre-stressed cables which run through pre-fabricated holes in Osteomorphic blocks. It is shown that the pre-stressed steel cables can provide the necessary constraint force without creating appreciable residual stresses in the cables, however the points of connection of the cables are the weakest points and need special treatment. The main mechanical feature of the interlocking structures is the absence of block bonding. As a result, the blocks have a certain freedom of translational and rotational movement (within the kinematic constraints of the assembly) and their contacts have reduced shear stresses which hampers fracture propagation from one block to another. These features pre-determine the specific ways the interlocking assemblies behave under mechanical and dynamic impacts. These were studied in this project and the following results are reported. As the blocks in the interlocking structure are not connected, the main issue is the bearing capacity. The study of the least favourable, central point loading in the direction normal to the structure shows elevated large-scale fracture toughness (resistance to fracture propagation). However when the central force imposes considerable bending the generated tensile membrane stresses assist fracturing of the loaded block. Prevention of bending considerably enhances the strength therefore the most efficient application of the interlocking structures would be in protective coatings and covers. Furthermore, proper selection of the material properties and the interface friction can increase the system overall strength and bearing capacity. The results of the computer simulations suggest that both Young’s modulus and the friction coefficient are the key parameters whose increase improves the bearing capacity of topologically interlocking assemblies.

Materials have traditionally been selected for the design of a product; however, advances in the understanding of material processing along with simulation and computation techniques are now making it possible to systematically design materials by tailoring the properties of the material to achieve the desired product performance. Material design offers the potential to increase design freedom and enable improved product performance; however, this increase in design freedom brings with it significant complexity in predictive models used for design, as well as many new design variables to consider. Material selection, on the other hand, is a well-established method for identifying the best materials for a product and does not require the complex models needed for material design. But material selection inherently limits the design of products by only considering existing materials. To balance increasing design costs with potentially improved product performance, designers must have a method for assessing the value of material design in the context of product design. In this thesis, the Design Space Expansion Strategy (DSES) and the Value of Design Space Expansion (VDSE) metric are proposed for supporting a designer s decision between material selection and material design in the context of product design. The strategy consists of formulating and solving two compromise Decision Support Problems (cDSP). The first cDSP is formulated and solved using a selected baseline material. The second cDSP is formulated and solved in an expanded material design space defined by material property variables in addition to other system variables. The two design solutions are then compared using the VDSE metric to quantify the value of expanding the material design space. This strategy is demonstrated in this thesis with an example of blast resistant panel design and is validated by application of the validation square, a framework for the validating design methods.

A key challenge facing designers creating innovative products is concept generation. Conceptual design is more effective when the design space is broadened by using an integrated design of product and material concepts approach. Conceptual design can also be accelerated by including problem solving and solution triggering tools in its structure. In this approach, structured analogy is used to transfer underlying principles from a solution suitable in one domain (i.e., product or mechanical domain) to an analogous solution in another domain (i.e., material domain). The nature of design analogy does not require as full of an exploration of the target domain as would otherwise be necessary; affording the possibility of a more rapid development. The addition of problem solving and solution triggering tools to a design method also decreases the design time and/or improves the quality of the final solution. This approach is formulated through a combination of the Theory of Inventive Problem Solving (TRIZ) proposed by Altshuller, and the systematic approach of Pahl and Beitz, for products that are jointly considered at the product and material level. These types of problems are ones where customer performance requirements are fulfilled through both the designed product and the designed material. The systematic approach of Pahl and Beitz is used as the base method through which TRIZ is used as a means of transferring abstract information about the design problem between the domains with the aim of accelerating conceptual design. This also allows for multi-domain design tools such as Su-Field-Model integration with design repositories for the transfer of information at different levels of abstraction; expanding the design space and effectively directing the designer. The explanation of this approach is presented through a simple example of a spring design improvement and validated through concept generation of a reactive material containment system.

The material development cycle is becoming too slow if compared with other technologies sectors like IT and electronics. The materials scientists’ community needs to bring materials science back to the core of human development. ICME (Integrated Computational Materials Engineer) is a new discipline that uses advanced computational tools to simulate material microstructures, processes and their links with the final properties. There is the need for a new way to design tailor-made materials with a faster and cheaper development cycle while creating products that meet “real-world” functionalities rather than vague set of specifications. Using the ICME approach, cutting edge computational thermodynamics models were employed in order to assist the microstructure characterization and refinement during the TIG welding of a functionally graded composite material with outstanding wear and corrosion resistance. The DICTRA diffusion model accurately predicted the carbon diffusion during sintering, Thermo-Calc and TC-PRISMA models described the thermodynamic and kinetics of harmful carbide precipitation, while COMSOL Multhiphysic furnished the temperature distribution profile at every timestep during TIG welding of the material. Bainite transformation and the influence of chromium and molybdenum was studied and modelled with MAP_STEEL software. The simulations were then compared with experimental observations and a very good agreement between computational works and experiments was found for both thermodynamic and kinetics predictions. The use of this new system proved to be a robust assistance to the classic development method and the material microstructures and processes were carefully adjusted in order to increase corrosion resistance and weldability. This new approach to material development can radically change the way we think and we make materials. The results suggest that the use of computational tools is a reality that can dramatically increase the efficiency of the material development.

Designers are challenged to manage customer, technology, and socio-economic uncertainty causing dynamic, unquenchable demands on limited resources. In this context, increased concept flexibility, referring to a designer s ability to generate concepts, is crucial. Concept flexibility can be significantly increased through the integrated design of product and material concepts. Hence, the challenge is to leverage knowledge of material structure-property relations that significantly affect system concepts for function-based, systematic design of product and materials concepts in an integrated fashion. However, having selected an integrated product and material system concept, managing complexity in embodiment design-processes is important. Facing a complex network of decisions and evolving analysis models a designer needs the flexibility to systematically generate and evaluate embodiment design-process alternatives. In order to address these challenges and respond to the primary research question of how to increase a designer s concept and design-process flexibility to enhance product creation in the conceptual and early embodiment design phases, the primary hypothesis in this dissertation is embodied as a systematic approach for integrated product, materials and design-process design. The systematic approach consists of two components i) a function-based, systematic approach to the integrated design of product and material concepts from a systems perspective, and ii) a systematic strategy to design-process generation and selection based on a decision-centric perspective and a value-of-information-based Process Performance Indicator. The systematic approach is validated using the validation-square approach that consists of theoretical and empirical validation. Empirical validation of the framework is carried out using various examples including: i) design of a reactive material containment system, and ii) design of an optoelectronic communication system.