Search

• Propose a meta-learning-based surrogate modelling (MLSM) framework for knowledge transfer. • Provide the definition of similar tasks and identify the scope of the framework. • Outline the applications to global sensitivity analysis, reliability analysis, and optimization. • Demonstrate the ability of the framework to transfer prior knowledge and show the computational efficiency. Surrogate modelling has emerged as a useful technique to study complex physical and engineering systems in various disciplines, especially for engineering analysis. Previous studies mostly focused on developing new surrogate models and/or applying existing surrogate models to practical problems. Despite the computational efficiency, the surrogate for a new task is often built from scratch and the knowledge gained from previous surrogate modelling for similar tasks is neglected. As the need for quickly modifying simulation models to reflect design changes has significantly increased, one potential solution is to utilize prior knowledge from surrogate modelling. In this study, a novel meta-learning-based surrogate modelling framework is presented. The framework includes two phases: a meta-training and a few-shot learning phase. A meta-model that represents a family of tasks and the adaptation of this model to a new task with few data points are the results of the first and second phase, respectively. The study specifies the scope of the framework by classifying similar tasks. Applications of the framework to global sensitivity analysis, optimization, and reliability analysis are also addressed. Four numerical experiments are performed to demonstrate the feasibility and applicability of the framework. [ABSTRACT FROM AUTHOR]

• The paper presents a semi-analytical uncertainty quantification framework for CFD simulations. • Bayesian quadrature is more efficient and accurate than the random sampling method. • A proper input transformation may be needed when using a simple kernel function in the GP model. • The parametric uncertainties may introduce large uncertainties within the results. To improve the safety, reliability, and performance of complex engineering systems, it is crucial to understand and quantify uncertainties. This paper presents a framework to non-intrusively and semi-analytically quantify the parametric uncertainty within CFD simulations using Bayesian quadrature (BQ). An in-house uncertainty quantification (UQ) code based upon this mathematical framework is developed. The code is then validated by applying it to quantify the uncertainty due to a varying parameter in a simple analytical test function. The mean and variance obtained using BQ are compared with those obtained from the analytical solution and stochastic simulation using the Latin hypercube sampling (LHS) method. The validation test case shows that BQ outperforms the LHS approach in terms of computational efficiency and accuracy. The UQ code is then utilised to characterise the uncertainty (due to the unknown inlet flow profile) of CFD predicted operating parameters of an industrial scale butterfly valve, as well as the uncertainties (due to the unknown high-wavenumber damping factor C s) of a SAS-SST simulated bluff-body flow. It is found that the entry flow profile presents non-ignorable effects on the valve operating parameters. Meanwhile, the variance of the valve operating parameters changes with the valve opening. For the bluff-body flow, large variances of predicted flow properties exist in the region where the separate shear layer dominates because of varying C s. Moreover, the effect of C s is more significant on the turbulence quantities, as it acts on the generation of turbulent eddies directly. [ABSTRACT FROM AUTHOR]

A procedure for rapid EM-based multi-objective optimization of compact microwave components is presented. Our methodology employs a recently developed nested kriging modelling to identify the search space region containing the Pareto-optimal designs, and to construct a fast surrogate model. The latter permits determination of the initial Pareto set, further refined using a separate surrogate-assisted process. As an illustration, a three-section impedance transformer is designed for the best matching and minimum size. The set of trade-off designs is produced at the low computational cost of only a few hundred of high-fidelity EM simulations of the transformer circuit despite a large number of its geometry parameters. [ABSTRACT FROM AUTHOR]

The selection of stationary or non-stationary Kriging to create a surrogate model of a black box function requires apriori knowledge of the nature of response of the function as these techniques are better at representing some types of responses than others. While an adaptive technique has been previously proposed to adjust the level of stationarity within the surrogate model such a model can be prohibitively expensive to construct for high dimensional problems. An alternative approach is to employ a surrogate model constructed from an ensemble of stationary and non-stationary Kriging models. The following paper assesses the accuracy and optimization performance of such a modelling strategy using a number of analytical functions and engineering design problems. [ABSTRACT FROM AUTHOR]

Myriad natural protective structures consist of bone plates joined by convoluted unmineralized (soft) collagenous sutures. Examples of such protective structures include: shells of turtles, craniums of almost all animals (including humans), alligator armour, armadillo armour, and others. The function of sutures has been well researched. However, whether, and if so how, sutures improve protective performance during a predator attack has received limited attention. Sutures are ubiquitous in protective structures, and this motivates the question as to whether sutures optimize the protective function of the structure. Hence, in this work the behaviour of structures that contain sutures during predator attacks is investigated. We show that sutures decrease the maximum strain energy density that turtle shells experience during predator attacks by more than an order of magnitude. Hence, sutures make turtle shells far more resilient to material failure, such as, fracture, damage, and plastic deformations. Additionally, sutures increase the viscous behaviour of the shell causing increased dissipation of energy during predator attacks. Further investigations into the influence of sutures on behaviour during locomotion and breathing are also presented. The results presented in this work motivate the inclusion of sutures in biomimetically designed protective structures, such as helmets and protective clothing. STATEMENT OF SIGNIFICANCE: Myriad bony protective structures contain networks of sutures, that is con- voluted soft collagenous tissue. Their ubiquity motivates the question, whether, and if so how, sutures improve protective performance. Hence, this work inves- tigates how sutures affect protective performance using computational experi- ments. Due to the length scale of sutures being far smaller than the structures in which they reside, classical modelling approaches are prohibitively expensive. Hence, in this work, a multiscale approach is taken. To our knowledge, this is the first multiscale investigation of structures that contain sutures. Among other insights, we show that sutures decrease the maximum strain energy density in structures during predator attacks by over an order of mag- nitude. Hence, sutures make structures far more resilient to failure., Competing Interests: Declaration of Competing Interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper., (Copyright © 2022 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.)

Abstract: A new optimization algorithm is proposed and tested for four different test cases: benchmark test functions, multi-element airfoil optimization in subsonic flow, active flow control parameter optimization in transonic flow, and a direct shape optimization of an airfoil in transonic flow. The new algorithm emphasizes the use of an aerodynamic design prediction based on a global and a local surrogate modelling in genetic algorithm structure. A global response surface approximation is modelled by using low-order polynomials. A local response surface approximation is constructed by using neural networks. For all the demonstration problems considered herein, remarkable reductions in the computational times have been accomplished. The new approach significantly decreases the required CFD calls by approximately 50% in multi-element airfoil optimization; more than 70% in active flow control parameter optimization; and approximately 50% in direct shape optimization problem. [Copyright &y& Elsevier]

Abstract: Night cooling, especially in offices, attracts growing interest. Unfortunately, building designers face considerable problems with the case-specific convective heat transfer by night. The multizone building energy simulation programs they use actually need extra input, from either costly experiments or computational fluid dynamics (CFD) simulations. Alternatively, up-front research on how to engineer best night cooled spaces can thrust the application of night cooling. The authors of underlying paper set up a global surrogate-based optimization procedure to find room/system design solutions which induce a high convective heat flux during night cooling in a generic open plan office. This fully-automated configuration of data sampling, geometry/grid generation, CFD solving and surrogate modelling generated several surrogate models. These surrogate models indicated how the convective heat flux in the night cooled open plan office related to the ventilation concept, the thermal mass distribution, the geometry and the driving force for convective heat transfer. Actually these surrogate models merely guided the data sampling towards the global optimum. However, they also provided additional rough-hewn insights into the global behaviour. The results indicated that cases with thermal mass at the floor produce convective heat fluxes which are significantly higher than the ones with thermal mass at the ceiling. Among these cases, the performance of cross ventilation surpasses the ones of both single sided ventilation and under floor ventilation. Among the cases with a thermally massive ceiling, single sided ventilation seems superior. However, while cross ventilation is generally a robust ventilation concept, single sided ventilation is particularly sensitive to the geometry. [Copyright &y& Elsevier]

Abstract: When analysing data from computationally expensive simulation codes or process measurements, surrogate modelling methods are firmly established as facilitators for design space exploration, sensitivity analysis, visualisation and optimisation. Kriging is a popular surrogate modelling technique for data based on deterministic computer experiments. There exist several types of Kriging, mostly differing in the type of regression function used. Recently a promising new variable selection technique was proposed to identify a regression function in the Kriging framework. In this paper this type of Kriging, i.e., blind Kriging, has been efficiently implemented in Matlab® and has been extended. The implementation is validated and tested on several examples to illustrate the strength and weaknesses of this new, promising modelling technique. It is shown that the performance of blind Kriging is as good as, or better than ordinary Kriging. Though, blind Kriging comes at double the computational cost with respect to ordinary Kriging. [Copyright &y& Elsevier]

A prominent advantage of using surrogate models in structural design optimization is that computational effort can be greatly reduced without significantly compromising model accuracy. The essential goal is to perform the design optimization with fewer evaluations of the typically finite element analysis and ensuring accuracy of the optimization results. An adaptive surrogate based design optimization framework is proposed, in which Latin hypercube sampling and Kriging are used to build surrogate models. Accuracy of the models is improved adaptively using an infill criterion called expected improvement (EI). It is the anticipated improvement that an interpolation point will lead to the current surrogate models. The point that will lead to the maximum EI is searched and used as infill points at each iteration. For constrained optimization problems, the surrogate of constraint is also utilized to form a constrained EI as the corresponding infill criterion. Computational trials on mathematical test functions and on a three-dimensional aircraft wing model are carried out to test the feasibility of this method. Compared with the traditional surrogate base design optimization and direct optimization methods, this method can find the optimum design with fewer evaluations of the original system model and maintain good accuracy. [ABSTRACT FROM PUBLISHER]

Traditional surrogate modelling techniques, such as kriging, have been employed quite effectively within design optimizations. However, such models can fail to reproduce non-stationary responses accurately. This article explores the application of non-stationary kriging to design optimization and attempts to determine its applicability with regard to the optimization of both stationary and non-stationary objective functions. A series of analytical test problems and an engineering design problem are used to compare the performance of non-stationary and adaptive partial non-stationary kriging to traditional stationary kriging. [ABSTRACT FROM AUTHOR]

Abstract: We present here a multi-objective and multi-disciplinary coronary stent design optimization paradigm. Coronary stents are tubular, often mesh-like, structures which are deployed in diseased (stenosed) artery segments to provide a scaffolding feature that compresses atheromatus plaque, hence restoring luminal area and maintaining vessel patency. A three variable geometry parameterisation of a CYPHER (Cordis Corporation, Johnson & Johnson co.) type stent is proposed to explore the functionality of a sequence of circumferential rings connected by ‘n’ shaped links. The performance of each design is measured by six figures of merit (objectives/metrics) representing (i) acute recoil, (ii) tissue stresses, (iii) haemodynamic disturbance, (iv) drug delivery, (v) uniformity of drug distribution, and (vi) flexibility. These metrics are obtained from computational simulations of (i) structural deformation through balloon inflated expansion of a stent into contact with a stenosed vessel, (ii) pulsatile flow over the deformed stent embedded in the vessel wall, (iii) steady-state drug distribution into the tissue, and (iv) flexibility of a stent in response to an applied moment. Design improvement is obtained by a multi-objective surrogate modelling approach using a non-dominated sorting genetic algorithm (NSGA-II) to search for an optimal family of designs. A number of trade-offs between the different objectives are identified. In particular a conflict between pairs of the following objectives are shown -- (a) volume average stress vs. recoil, (b) volume average drug vs. volume average stress, (c) flexibility vs. volume average stress, (d) flexibility vs. haemodynamic disturbance, (e) volume average drug vs. haemodynamic disturbance, and (f) uniformity of drug vs. volume average stress. Different paradigms to choose the optimal designs from the obtained Pareto fronts are presented and under each such paradigm, the optimal designs and there relative positions with respect to a representative CYPHER stent are shown. The methodology and the results of this work could potentially be useful in further optimisation studies and development of a family of stents with increased resistance to in-stent restenosis and thrombosis. [Copyright &y& Elsevier]

In this article a new algorithm for optimization of multi-modal, nonlinear, black-box objective functions is introduced. It extends the recently-introduced adaptive multi-modal optimization by incorporating surrogate modelling features similar to response surface methods. The resulting algorithm has reduced computational intensity and is well-suited for optimization of expensive objective functions. It relies on an adaptive, multi-resolution mesh to obtain an initial estimation of the objective function surface. Local surrogate models are then constructed and used to generate additional trial points around the local minima discovered. The steps of mesh refinement and surrogate modelling continue until convergence is achieved. The algorithm produces progressively accurate surrogate models, which can be used for post-optimization studies such as sensitivity and tolerance analyses with minimal computational effort. This article demonstrates the effectiveness of the algorithm using comparative optimization of several multi-modal objective functions, and shows an engineering application of the design of a power electronic converter. [ABSTRACT FROM AUTHOR]

The dielectric barrier discharge (DBD) plasma actuator offers promising opportunities for flow control because of its fast response and non-moving parts. In this work, surrogate modelling is adopted to better understand the impact of the materials and operational parameters on the actuator performance, and to provide an efficient approach for performance estimation. The DBD model based on 2-species helium chemistry engages three design variables: operating frequency, polarity (positive/negative) time ratio of the applied voltage and the dielectric constant of the insulator. Two objectives are identified: the net force generated and the power requirement. Multiple surrogate models are used, which identify two branches of the Pareto front with opposite net force direction and substantially different parametric sensitivities. Global sensitivity analysis indicates that the voltage frequency and polarity ratio are important in different regions of the design space, while the dielectric constant is always important. [ABSTRACT FROM AUTHOR]

Abstract: Computational fluids dynamics (CFDs) and numerical optimization techniques are applied in an integrated methodology to explore the effects of different geometric characteristics on fluid mixing in a staggered herringbone micromixer (SHM). To quantify the mixing intensity in the mixer a mixing index is defined on the basis of the intensity of segregation of the mass concentration on a cross-section plane in the mixing channel. Four geometric parameters, i.e., aspect ratio of the mixing channel, ratio of groove depth to channel height, ratio of groove width to groove pitch and the asymmetry factor (offset) of groove, are the design variables initially selected for optimization, then two more parameters, i.e., angle of the groove and number of grooves per channel section, are evaluated. The whole optimization is conducted with a multi-objective approach for which the mixing index at the outlet section and the pressure drop in the mixing channel are the performance criteria used as objective functions. The Pareto front of designs with the optimum trade-off, maximum mixing index with minimum pressure drop, is obtained. [Copyright &y& Elsevier]

Simulation models are now available to represent the sewer network, wastewater treatment plant and receiving water as an integrated system. These models can be combined with optimisation methods to improve overall system performance through optimal control. Evolutionary algorithms (EAs) have been proven to be a powerful method in developing optimal control strategies; however, the intensive computational requirement of these methods imposes a limit on their application. This paper explores the potential of surrogate modelling in multiobjective optimisation of urban wastewater systems with a limited number of model simulations. A surrogate based method, ParEGO, is combined with an integrated urban wastewater model to solve real time control problems. This method is compared with the popular NSGA II, by using performance indicators: the hypervolume indicator, additive binary 1-indicator and attainment surface. Comparative results show that ParEGO is an efficient and effective method in deriving optimal control strategies for multiple objective control problems with a small number of simulations. It is suggested that ParEGO can greatly improve computational efficiency in the multiobjective optimisation process, particularly for complex urban wastewater systems. [ABSTRACT FROM AUTHOR]

Abstract: The design of water distribution networks is a large-scale combinatorial, non-linear optimisation problem, involving many complex implicit constraint sets, such as nodal mass balance and energy conservation, which are commonly satisfied through the use of hydraulic network solvers. These problem properties have motivated several prior studies to use stochastic search optimisation, because these derivative-free global search algorithms have been shown to obtain higher quality solutions for large network design problems. Global stochastic search methods, however, require many iterations to be performed in order to achieve a satisfactory solution, and each iteration may involve running computationally expensive simulations. Recently, this problem has been compounded by the evident need to embrace more than a single measure of performance into the design process, since by nature multi-objective optimisation methods require even more iterations. The use of metamodels as surrogates for the expensive simulation functions has been investigated as a possible remedy to this problem. However, the identification of reliable surrogates is not always a viable alternative. Under these circumstances, methods that are capable of achieving a satisfactory level of performance with a limited number of function evaluations represent a valuable alternative. This paper represents a first step towards filling this gap. Two recently introduced multi-objective, hybrid algorithms, ParEGO and LEMMO, are tested on the design problem of a real medium-size network in Southern Italy, and a real large-size network in the UK under a scenario of a severely restricted number of function evaluations. The results obtained suggest that the use of both algorithms, in particular LEMMO, could be successfully extended to the efficient design of large-scale water distribution networks. [Copyright &y& Elsevier]

• An adaptive reliability method for multi-fidelity models called AMGPRA is proposed. • Collective learning function (CLF) is proposed for characterizing the global impact of candidate points. • CLF simultaneously identifies the optimal training point and information source. • CLF can incorporate different core learning functions to adapt to different problems. • AMGPRA is applied to benchmark problems and wind reliability of transmission towers. In many fields of science and engineering, models with different fidelities are available. Physical experiments or detailed simulations that accurately capture the behavior of the system are regarded as high-fidelity models with low model uncertainty; however, they are expensive to run. On the other hand, simplified physical experiments or numerical models are seen as low-fidelity models that are cheaper to evaluate. Although low-fidelity models are often not suitable for direct use in reliability analysis due to their low accuracy, they can offer information about the trend of the high-fidelity model thus providing the opportunity to explore the design space at a low cost. This study presents a new approach called adaptive multi-fidelity Gaussian process for reliability analysis (AMGPRA). Contrary to selecting training points and information sources in two separate stages as done in state-of-the-art mfEGRA method, the proposed approach finds the optimal training point and information source simultaneously using the novel collective learning function (CLF). CLF is able to assess the global impact of a candidate training point from an information source and it accommodates any learning function that satisfies a certain profile. In this context, CLF provides a new direction for quantifying the impact of new training points and can be easily extended with new learning functions to adapt to different reliability problems. The performance of the proposed method is demonstrated by three mathematical examples and one engineering problem concerning the wind reliability of transmission towers. It is shown that the proposed method achieves similar or higher accuracy with reduced computational costs compared to state-of-the-art single and multi-fidelity methods. A key application of AMGPRA is high-fidelity fragility modeling using complex and costly physics-based computational models. [ABSTRACT FROM AUTHOR]

• Imprecise probabilities allow for explicitly taking epistemic uncertainty into account. • P-boxes are of interest for engineering analysis due to their mathematical straightforward description. • An overview of methods for construction of p-boxes based on available information is given. • The state-of-the-art in propagation schemes for p-boxes is discussed. The consideration of imprecise probability in engineering analysis to account for missing, vague or incomplete data in the description of model uncertainties is a fast-growing field of research. Probability-boxes (p-boxes) are of particular interest in an engineering context, since they offer a mathematically straightforward description of imprecise probabilities, as well as allow for an intuitive visualisation. In essence, p-boxes are defined via lower and upper bounds on the cumulative distribution function of a random variable whose exact probability distribution is unknown. However, the propagation of p-boxes on model inputs towards bounds on probabilistic measures describing the uncertainty on the model responses is numerically still very demanding, and hence is subject of intensive research. In order to provide an overview on the available methods, this paper gives a state-of-the art review for the modelling and propagation of p-boxes with a special focus on structural reliability analysis. [ABSTRACT FROM AUTHOR]

Optimization can be a time-consuming and demanding process when an analytical model of the system of interest cannot be developed. The problem is even more extreme when the robustification of the system is desired. Under such circumstances the design engineer will typically resort to design of experiments (DOE) or finite element analysis/computational fluid dynamics or some combination. Each can be time consuming and demanding. It is shown in this paper that by combining dimensional analysis with DOE it is possible to generate a near-exact surrogate model of a system empirically. This can be done with a significantly reduced number of experiments when compared with traditional DOE techniques. The approximation is sufficiently accurate to be optimized or robustified using methods traditionally suited to analytical models. A strategy designed to help the design engineer take full advantage of this approach is presented. [ABSTRACT FROM AUTHOR]

The following work documents the development of an artificial neural network (ANN) model of an advanced, turbocharged diesel engine, and its implementation within an overall vehicle simulation. The purpose of this research was to create a model that would help drastically reduce in-vehicle engine design cycle time, improve numerical robustness and maintain the predictive capability of its physically based counterpart. Ease of integration within an optimization framework for preliminary investigations of engine designs was also a significant need. A design of experiments (DOE) was first used to sample the engine's feasible design domain. The high-fidelity model calculated engine responses at these DOE designs; the DOE inputs and engine model responses were subsequently used to train a radial basis function ANN. A typical, vehicle-level optimization study is presented which compares results and time requirements of the ANN and high-fidelity models. Critical issues in the development of an ANN engine model are detailed and generic guidelines are presented to expedite the implementation of this surrogate modelling technique to other complex subsystems and components. [ABSTRACT FROM AUTHOR]

The space mapping technique is intended for optimization of engineering models which involve very expensive function evaluations. It is assumed that two different models of the same physical system are available: Besides the expensive model of primary interest (denoted the fine model), access to a cheaper (coarse) model is assumed which may be less accurate. The main idea of the space mapping technique is to use the coarse model to gain information about the fine model, and to apply this in the search for an optimal solution of the latter. Thus the technique iteratively establishes a mapping between the parameters of the two models which relate similar model responses. Having this mapping, most of the model evaluations can be directed to the fast coarse model. In many cases this technique quickly provides an approximate optimal solution to the fine model that is sufficiently accurate for engineering purposes. Thus the space mapping technique may be considered a preprocessing technique that perhaps must be succeeded by use of classical optimization techniques. We present an automatic scheme which integrates the space mapping and classical techniques. [ABSTRACT FROM AUTHOR]

In-stent restenosis is a recurrence of coronary artery narrowing due to vascular injury caused by balloon dilation and stent placement. It may lead to the relapse of angina symptoms or to an acute coronary syndrome. An uncertainty quantification of a model for in-stent restenosis with four uncertain parameters (endothelium regeneration time, the threshold strain for smooth muscle cell bond breaking, blood flow velocity and the percentage of fenestration in the internal elastic lamina) is presented. Two quantities of interest were studied, namely the average cross-sectional area and the maximum relative area loss in a vessel. Owing to the high computational cost required for uncertainty quantification, a surrogate model, based on Gaussian process regression with proper orthogonal decomposition, was developed and subsequently used for model response evaluation in the uncertainty quantification. A detailed analysis of the uncertainty propagation is presented. Around 11% and 16% uncertainty is observed on the two quantities of interest, respectively, and the uncertainty estimates show that a higher fenestration mainly determines the uncertainty in the neointimal growth at the initial stage of the process. The uncertainties in blood flow velocity and endothelium regeneration time mainly determine the uncertainty in the quantities of interest at the later, clinically relevant stages of the restenosis process.

• Apply directional sampling to auto-generate an inventory of 720 RCMRF office building of varying topology, and use performance-based earthquake engineering methods to assess economic and environmental impacts associated with seismic hazard in Charleston, SC. • Leverage inventory to build a machine learning (ML) pipeline and compare five common ML models as surrogate for life cycle resilience and sustainability impacts using only input parameters known/estimable during early design. • Analyze the sensitivity of models to identify the most influential parameters and assess tradeoffs associated with using only crude topological parameters. • Support vector regression found to achieve an acceptable balance of accuracy, interpretability and generalizability, even when using only topological parameters and can serve as an aid to structural engineers in early design. This paper presents a framework to develop generalizable surrogate models to predict seismic vulnerability and environmental impacts of a class of buildings at a particular location. To this end, surrogate models are trained on a performance inventory, here simulation-based seismic and environmental assessments of 720 mid-rise concrete office buildings of variable topology in Charleston, South Carolina. Five surrogate models of multiple regression, random forest, extreme gradient boosting, support vector machine and k-nearest neighbors were trained in a machine-learning pipeline including hyperparameter tuning and cross-validation. Variance-based sensitivity and accumulated local effect analysis were performed on the most accurate model to identify the most influential parameters and interpret the trained surrogate model. Support vector machines achieved the highest accuracy for total annual loss with an average 10-fold adjusted R2 of 0.96, whereas simpler linear regression was adequate to estimate the initial and seismic-induced embodied carbon emission. Floor area, building height, lateral-resisting frame weight, and average beam section sizes were found to be the most influential features. As these features may be approximated by an experienced structural engineer the results indicate that, with suitable performance inventories available, it should be possible to employ surrogate models in early design to narrow the initial design space to highly resilient and sustainable configurations. [ABSTRACT FROM AUTHOR]

The study focuses on both the effects of soil foundation liquefaction on the induced damage of an embankment and on the generation of a probabilistic seismic demand model that relates few specific characteristics of seismic hazard at a site with the liquefaction-induced settlement. To this aim, several non-linear dynamic Finite Element (FE) analyses were performed. However, given the costly evaluation of the numerical model simulations for a large number of earthquake records, a surrogate model based on multi-fidelity approach is used to represent the output of the expensive FE model. In this approach, the training database is composed by computational low-fidelity data together with limited high-fidelity one. Thus, the proposed methodology is used to generate the curves describing the annual rate of exceeding different values of damage levels. A comparison with the FE reference results suggests that the accuracy of the prediction of the proposed surrogate model is comparable to those from direct numerical FE analysis. Findings also illustrate clearly the importance and the advantages of an adequate definition of the input parameters to build the surrogate model. • A Probabilistic seismic demand model relating few specific characteristics of seismic hazard at a site with the liquefaction-induced settlement. • The model overcomes the high computational cost due to mathematical complexities of the numerical model. • A model easily to implement with a great potential to be used in engineering practice. • The model assumptions are flexible enough to accommodate various modelling needs. • The model allows to generate the curves describing the annual rate of exceeding different values of damage levels. [ABSTRACT FROM AUTHOR]

We present a Physics-Informed Neural Network (PINN) to simulate the thermochemical evolution of a composite material on a tool undergoing cure in an autoclave. In particular, we solve the governing coupled system of differential equations – including conductive heat transfer and resin cure kinetics – by optimizing the parameters of a deep neural network (DNN) using a physics-based loss function. To account for the vastly different behaviour of thermal conduction and resin cure, we design a PINN consisting of two disconnected subnetworks, and develop a sequential training algorithm that mitigates instability present in traditional training methods. Further, we incorporate explicit discontinuities into the DNN at the composite-tool interface and enforce known physical behaviour directly in the loss function to improve the solution near the interface. We train the PINN with a technique that automatically adapts the weights on the loss terms corresponding to PDE, boundary, interface, and initial conditions. Finally, we demonstrate that one can include problem parameters as an input to the model – resulting in a surrogate that provides real-time simulation for a range of problem settings – and that one can use transfer learning to significantly reduce the training time for problem settings similar to that of an initial trained model. The performance of the proposed PINN is demonstrated in multiple scenarios with different material thicknesses and thermal boundary conditions. • A physics-informed neural network modelling thermochemical changes in materials. • Solving coupled differential equations for multiphysics event in bi-material system. • A sequential training algorithm on disconnected networks to mitigate instability. • Enforcing physical behaviour at the designed discontinuities in the neural network. [ABSTRACT FROM AUTHOR]

The In-Stent Restenosis 2D model is a full y coupled multiscale simulation of post-stenting tissue growth, in which the most costly submodel is the blood flow simulation. This paper presents uncertainty estimations of the response of this model, as obtained by both non-intrusive and semi-intrusive uncertainty quantification. A surrogate model based on Gaussian process regression for non-intrusive uncertainty quantification takes the whole model as a black-box and maps directly the three uncertain inputs to the quantity of interest, the neointimal area. The corresponding uncertain estimates matched the results from quasi-Monte Carlo simulations well. In the semi-intrusive uncertainty quantification, the most expensive submodel is replaced with a surrogate model. We developed a surrogate model for the blood flow simulation by using a convolutional neural network. The semi-intrusive method with the new surrogate model offered efficient estimates of uncertainty and sensitivity while keeping a relatively high accuracy. It outperformed the results obtained with earlier surrogate models. It also achieved the estimates comparable to the non-intrusive method with a similar efficiency. Presented results on uncertainty propagation with non-intrusive and semi-intrusive metamodelling methods allow us to draw some conclusions on the advantages and limitations of these methods. • Uncertainty quantification analysis is carried out for a 2D in-stent restenosis model. • Non- and semi-intrusive methods with a surrogate model are applied in the experiments. • Both methods work well compared to the result from the quasi-Monte Carlo method. • The properties of the methods are studied and discussed. [ABSTRACT FROM AUTHOR]

Electromagnetic simulation tools have been playing an increasing role in the design of contemporary antenna structures. The employment of electromagnetic analysis ensures reliability of evaluating antenna characteristics but also incurs considerable computational expenses whenever massive simulations are involved (e.g., parametric optimization, uncertainty quantification). This high cost is the most serious bottleneck of simulation-driven design procedures, and may be troublesome even for local tuning of geometry parameters, let alone global optimization. On the one hand, globalized search is often necessary because the design problem might be multimodal (i.e., the objective function features multiple local optima) or a reasonably good initial design may not be available. On the other hand, the computational efficiency of popular algorithmic approaches, primarily, nature-inspired population-based algorithms, is generally poor. Combining metaheuristics procedures with surrogate modelling techniques and sequential sampling methods alleviates the problem to a certain extent but modelling of nonlinear antenna responses over broad frequency ranges is extremely challenging, and the aforementioned solutions are normally limited to rather simple structures described by a few parameters. This paper proposes a novel approach to global optimization of multi-band antennas. The major component of the presented framework is the knowledge-based inverse surrogate constructed at the level of response features (e.g., frequency and level locations of the antenna resonances). The surrogate facilitates decision-making process of inexpensive identification of the most promising regions of the parameter space, and a rendition of the good-quality initial design for further local tuning. Our methodology is validated using three examples of dual- and triple-band antennas. The average optimization cost is only 150 full-wave antenna analyses while ensuring precise allocation of the antenna resonances at the target frequencies. This performance is demonstrated superior over both local optimizers and population-based metaheuristics. [ABSTRACT FROM AUTHOR]