Due to high seismic vulnerability and severity of possible failure consequences, petrochemical installations are often considered as “special risk” plants. Although tanks, pipes, elbows and bolted flanges have been a major concern in terms of seismic design, generally, they have not been analysed with modern performance-based procedures. This thesis will explore some important themes in seismic risk assessment with a special focus on petrochemical plants and components. In the first part of the thesis the case study of a probabilistic seismic demand analysis (PSDA) for a Refrigerated liquefied gas (RLG) subplant is presented. As a matter of fact, RLG terminals that are part of strategic facilities must be able to withstand extreme earthquakes. In detail, a liquefied natural gas (LNG, ethylene) terminal consists of a series of process facilities connected by pipelines of various sizes. In this study, the seismic performance of pipes, elbows and bolted flanges is assessed, and seismic fragility functions are presented within the performance-based earthquake engineering framework. Particular attention is paid to component resistance to leakage and loss of containment (LoC) even though several different limit states are investigated. The LNG tank, support structures and pipework, including elbows and flanges, are analysed with a detailed 3D finite element model. For this purpose, a mechanical model of bolted flange joints is developed, able to predict the leakage limit state, based on experimental data. A significant effort is also devoted to identification of a leakage limit state for piping elbows, and the level of hoop plastic strain was found to be an indicator. The second part of the thesis describes an innovative methodology to evaluate seismic performances of a realistic tank-piping system with special focus on LoC from piping elbows. This methodology relies on a set of experimental dynamic tests performed throughout hybrid simulations where the steel storage tank is numerically modelled while, conversely, the physical substructure encompasses the coupled piping network. Besides, ground motions for dynamic tests are synthetized based on a stochastic ground motion model whose input parameters are derived from the results provided by a seismic hazard analysis. Then, based on output data from the experimental tests, both a high-fidelity and a low-fidelity FE model are calibrated. Furthermore, these models are used to run additional seismic analyses using a large set of synthetic ground motions. Moreover, in order to derive the seismic response directly from inputs parameters of the stochastic ground motions model, the procedure to build a hierarchical kriging surrogate model of the tank-piping system is presented. Eventually, the surrogate model can be adopted to perform a seismic fragility analysis. Along with the line of probabilistic analysis, another contribution to this research work is a probabilistic seismic demand model (PSDM) of a steel-concrete composite structure made of a novel type of high-strength steel moment resisting frame. According to the main topic of this thesis, the procedure that is here presented can be used either in a seismic risk assessment or a fully probabilistic performance-based earthquake engineering (PBEE) framework. In detail a 3D probabilistic seismic demand analysis was performed considering the variability of the earthquake incident angle, generally not taken in account in typical fragility analyses. Therefore, the fragility curves evaluated following this approach account for the uncertainty of both the seismic action and its direction.

The consistent, uniform pressing of green bodies is a necessary part of producing high-quality, high-performance ceramics with predictable qualities and behavior. Undesirable density variation in the compacted ceramic powder causes variability in performance, failure to meet quality control standards, and, possibly, complete piece failure during successive processing. These issues contribute directly to a decrease in production efficiency through lost time and an increase in energy and material use. The careful control of the green body density field is of the utmost importance to consistently producing high-performance ceramics. Current methods for minimizing heterogeneity of the density field are often based on trial-and-error to optimize mold geometry and forming pressure, which is both expensive and prolongs development. The present research presents a continuum-level constitutive model for accurately modeling the densification of ceramic powders into green bodies and outlines the numerical implimentation of said model. The constitutive model incorporates nonlinear elasticity, elatic-plastic coupling, cap evolution, pressure- and Lode angle-dependent plasticity, and hardening. To evaluate the constitutive model, a new method for measuring density in green bodies has been developed. This method utilizes readily-available laboratory equipment to produce density projection data for the sample and subsequently processes that data to produce a 3D density field using well-developed tomographic reconstruction techniques. Finally, a green body is produced from alumina powder (Martoxid KMS-96) and the density field is evaluated and compared to that of a numerical simulation. They are shown to agree within the error of the density measurements. These comparisons demonstrate the performance of the developed constitutive model and the potential utility for companies and research institutions that are in the ceramics production field.

The mechanical behaviour of wood is highly conditioned by the duration of load with regard to both its resistance and stiffness. In particular the problem of the deformability of floors is an essential aspect in design of wooden structures since, as to ordinary span beams, the service behaviour is generally a more severe standard of verification than the ultimate limit state. This research concerned the investigation of the rheological behaviour of timber and timber-concrete composite beams through an experimental approach, in order to go deep into the different components that influence the development of the long term deflection: bending and shear creep, as well as creep of connection for composite beams. The addition of a collaborating reinforced concrete slab is an ordinary solution for improving the stiffness and resistance of wooden beams, the validity of which depends on the deformability of the connection system, in addition to the mechanical properties of timber and concrete. In this work was investigated the long term behaviour of an 8 meter span timber-concrete composite beam, with stud connection. The test was carried out indoor, in variable hygrometric conditions and the results were compared with the forecasts of the method suggested by Eurocode 5. A second area of investigation concerned the rheological behaviour of the wood subject to shear, which represents one of the less explored aspects of the research into the wood. That behaviour has been investigated through long term tests on small specimens subject to different stress levels, performed indoor at uniform temperature and variable environmental relative humidity. In that case too, the results were compared with the previsions of the codes and rheological model proposed by Toratti. By experimentally calibrating the parameters of that model was made a forecast of the creep coefficient for the service life of the structures in service class 1 and 3. The inquiry into the phenomena that cause the rheological behaviour of timber and composite beams was completed through bending tests on wooden beams and shear tests on timber-concrete and timber-lime mortar stud connections, carried out in the same ambient where the test on the composite beam was made. ------------------------------------------------------------ Il comportamento meccanico del legno è fortemente condizionato dalla durata del carico, sia in termini di resistenza, sia in termini di rigidezza. Il problema della deformabilità degli impalcati, in particolare, rappresenta un aspetto essenziale nella progettazione delle strutture in legno, poiché, per le travi di luce ordinaria, il comportamento in esercizio costituisce in genere un criterio di verifica più severo rispetto allo stato limite ultimo. Il presente lavoro di ricerca ha riguardato lo studio del comportamento reologico delle travi in legno e delle travi miste legno-calcestruzzo, mediante un approccio sperimentale teso ad approfondire le diverse componenti che contribuiscono a caratterizzare lo sviluppo della deformazione differita: la viscosità a flessione, a taglio e, nella trave mista, la viscosità della connessione. L’accoppiamento con una lastra collaborante in calcestruzzo armato rappresenta una soluzione diffusa per il miglioramento della rigidezza, oltre che della resistenza, delle travi in legno, la cui validità dipende dall’efficacia della connessione tra i due materiali, oltre che dalle caratteristiche di questi ultimi. In questa sede è stato indagato, mediante una prova di flessione di lunga durata, il comportamento deformativo di una trave mista legno-calcestruzzo di 8 metri di luce con connessione a piolo. La prova è stata condotta in ambiente interno con condizioni igrometriche variabili e i risultati sono stati posti a confronto con le previsioni del metodo suggerito dall’Eurocodice 5. Un secondo ambito di ricerca ha riguardato il comportamento reologico del legno soggetto ad azione di taglio, che rappresenta uno degli aspetti meno esplorati nelle indagini sul legno. Tale comportamento è stato caratterizzato mediante prove di lunga durata su campioni di piccole dimensioni soggetti a diversi livelli di sforzo, in ambiente controllato a temperatura costante e umidità ambientale relativa variabile; anche in questo caso i risultati ottenuti sono stati posti a confronto con le previsioni della normativa e con il modello reologico proposto da Toratti. Calibrando sperimentalmente i parametri di questo modello, è stata effettuata una previsione del coefficiente di viscosità per la vita utile delle strutture in classe di servizio 1 e 3. L’indagine sui fenomeni che determinano il comportamento reologico delle travi in legno e delle travi miste è stata completata mediante prove di flessione su travi in legno e prove di taglio su connessioni a piolo legno-calcestruzzo e legno-malta di calce naturale, condotte nello stesso ambiente in cui si è svolta la prova sulla trave mista.

After-earthquake assessment of buildings in terms of usability and safety is nowadays performed by in-charge technicians which are called to give their judgment basing mainly on in-field surveys and visual inspections. This necessarily implies additional inconvenience for residents and economic losses in the affected area, being often large the time required for conducting the surveys and being the judgment on the safe side in absence of objective data. A near real-time assessment based on objective data related to the seismic response of the structures is possible though the use of a monitoring systems capable of providing information on the state of the monitored structure inferring observations of its dynamic response. One of the most reliable parameter which can be correlated to the state of condition of a structure after an earthquake is the ductility demand expressed in terms of interstory drift. The use in monitoring systems of this indicator is examined in this thesis through case studies on reinforced concrete framed buildings and precast industrial buildings. In the design process of the systems I proposed a capacity-demand approach, through the prior formal definition of the requirements of accuracy and the calculation of the actual accuracy of the designed monitoring system. In particular I investigated in detail the uncertainties, both instrumental and related to model, to be combined in order to obtain the overall uncertainty of the information provided by the monitoring system, when using the method of double integration of the acceleration measurements. I have found that in general the instrumental uncertainties have less importance to the uncertainties of the model, in particular in presence of residual displacements at the end of the seismic motion. Aiming to reduce uncertainties in the presence of residual displacements and to cancel the need of high-pass filtering acceleration signals, I proposed a sensing bar prototype instrumented with accelerometers and inclinometers.

Hybrid Simulation with Dynamic Substructuring (HSDS) is a mixed numerical/- experimental simulation techniques. In detail, HSDS combines a Physical Substructure(PS) -the most critical subpart- with a Numerical Substructure (NS), and a compliant time integration process calculates the overall dynamic response of the emulated system. With the objective to circumvent three among major limitations of HSDS, the present thesis offers methodological procedures and algorithms aimed at: i) emulating a consistent degradation between PSs and NSs via model updating techniques; ii) handling PSs characterized by several internal DoFs with a reduced number of interface actuation points; iii) improving the computational efficiency in the case of complex NSs via partitioned time integrators. An old reinforced concrete bridge and a steel piping network for industrial plants are introduced as full-scale structural case studies. Part of significant results were published on referee journals and proceedings of international conferences. Part of developed tools was uploaded to the NEESHub web repository that is a United States web platform for research, collaboration and education powered by the George E. Brown, Jr. Network for Earthquake Engineering Simulation (NEES).

Through a second-order homogenization procedure, the explicit relation is obtained between the non-local parameters of a second gradient elastic ma- terial and the microstructure of a composite material. This result is instru- mental for the definition of higher-order models, to be used for the analysis of mechanics at micro- and nano-scale, where size-effects become important. The obtained relation is valid for both plane and three-dimensional prob- lems and generalizes earlier findings by Bigoni and Drugan (Analytical deriva- tion of Cosserat moduli via homogenization of heterogeneous elastic materials. J. Appl. Mech., 2007, 74, 741753) from several points of view: i) the result holds for anisotropic phases with spherical or circular ellipsoid of inertia; ii) the displacement boundary conditions considered in the homogenization procedure is independent of the characteristics of the material; iii) a perfect energy match is found between heterogeneous and equivalent materials (instead of an optimal bound). From the obtained solution it follows that the equivalent second-gradient Mindlin elastic solid: a) is positive definite only when the discrepancy tensor is negative defined; b) the non-local material symmetries are the same of the discrepancy tensor; c) the non-local effective behaviour is affected by the shape of the RVE, which does not influence the first-order homogenized response. Finally, explicit derivations of non-local parameters from heterogeneous Cauchy elastic composites are obtained in particular cases.

An approach based on genetic algorithm and parallel computing has been presented and discussed for structural optimizations. Some details on its software implementation are given and explained. Numerical simulations demonstrate the applicability of the proposed approach for the optimization of large-scale real structures.

TThe present work addresses the rigorous derivation of the Flügge treatment of the buckling of a thin cylinder. The incremental equilibrium equations in terms of generalized stresses are rigorously derived in terms of mean quantities (holding true regardless of the thickness of the cylinder), through a generalization of the approach introduced by Biot (1965) for rectangular plates. The incremental kinematics is postulated through a novel deduction from the deformation of a two-dimensional surface, thus generalizing an approach introduced to derive the incremental kinematics of a plate. The nonlinear elastic constitutive equations proposed by Pence and Gou (2015), describing a nearly incompressible neo-Hookean material, are used in a rigorous way. While the employed kinematics coincides with that used by Flügge, the incremental equilibrium and constitutive equations derived in this work are different from those given by Flügge, but are shown to reduce to the latter by invoking the smallness of the cylinder wall. The equations derived for the incremental deformation of prestressed thin cylindrical shells are general and can be used for different purposes. The study of the bifurcation problem of a thin-walled circular cylinder subject to compressive load is offered. When compared, the bifurcation landscape obtained from the formulation developed in this work and that given by Flügge are numerically shown to coincide and be consistent with results obtained by a fully three-dimensional theory of nonlinear elasticity. Furthermore the formula for the axial buckling stress of a ‘mid-long’ cylindrical shell made of a nearly incompressible neo-Hookean material and of a Mooney-Rivlin material are rigorously obtained from the presented formulation.

This thesis regards the study of the seismic behaviour of timber-frame buildings. Three are the main sections. Firstly, the study of the linear and non-linear behaviour of a timber-frame wall subjected to a horizontal force is presented, suggesting some analytical expression to correlate the mechanical behaviour of the entire wall to the mechanical properties of connection devices (i.e. fasteners, angle brackets and hold-down). Particular attention was paid to the ductility of each component. Secondly, a numerical modelling for the seismic linear analysis of multi-storey walls is proposed. In this section the horizontal force distribution between the walls is investigated too. Thirdly, a full-scale shake table test on a prefabricated 3-storey timber-frame building is described.

Although industrial piping systems and their components have been found highly vulnerable under earthquake events, there exists an inadequacy of proper seismic analysis and design rules for these structures. Current seismic design Standards and Codes are found to be over-conservative and some components, e.g., elbows, bolted flange joints and Tee joints, do not have detailed design guidelines that take into account earthquake loading. Thus, a clear need for the development of improved seismic design rules for such systems is evident. In this respect, numerical and experimental studies on piping systems and their components subjected to earthquake loading could be useful. As a result, valuable information, such as seismic capacities and demands under different limit states, could be utilized for the amendment of relevant design Codes and Standards. This thesis undertook a numerical and experimental investigation on a typical industrial piping system and some of its components in order to assess their seismic performance. In particular, the following issues have been pursued: (i) design of two non-standard Bolted Flange Joints (BFJs) suitable for seismic applications; (ii) experimental testing of the designed BFJs under monotonic and cyclic loading in order to check their leakage, bending and axial capacities; (iii) finite element analysis of a piping system containing several critical components under seismic loading; (iv) implementation of a pseudo-dynamic and real time testing schemes to test the piping system under seismic loading; and (v) pseudo-dynamic and real time tests on the piping system under several levels of earthquake loading corresponding to both serviceability and ultimate limit states. The above-mentioned activities were attained in this thesis. In particular, two different non-standard BFJs, comparatively thinner than the Standard ones, were designed, and their performance was examined through a number of monotonic and cyclic tests. Experimental results exhibited a favourable performance of the BFJs under bending and axial loading and moderate internal pressure; a good capacity in terms of strength, ductility, energy dissipation and leakage was observed. Performance of a typical full-scale industrial piping system containing several critical components, such as elbows, a bolted flange joint and a Tee joint, under realistic seismic loading was investigated through extensive numerical and experimental activities. The techniques of pseudo-dynamic and real time testing with dynamic substructuring –hybrid testing- were adopted to carry out experimental activities on the piping system under several limit state earthquake loading suggested by performance-based earthquake Standards. Implementations of hybrid tests were challenging mainly because the piping system was endowed with distributed masses and subjected to distributed earthquake forces, for which these experimental techniques have been considered inadequate so far. A number of mode synthesis techniques, namely the Craig-Bampton and SEREP methods, were discussed and their effectiveness was analysed for the realization of these tests. A characterization of the actuators to be used in the experimental tests was performed based on a transfer function. Relevant hybrid tests were successfully executed and they displayed a favourable performance of the piping system and its components; they remained below yield limits without any leakage even for the collapse limit state.