Search results

The purpose of this paper is to develop surrogate models, using deep learning (DL), that can facilitate the application of EM analysis software. In the current status quo, electrical systems can be found in an ever-increasing range of products that are part of everyone’s daily live. With the advances in technology, industries such as the automotive, communications and medical devices have been disrupted with new electrical and electronic systems. The innovation and development of such systems with increasing complexity over time has been supported by the increased use of electromagnetic (EM) analysis software. Such software enables engineers to virtually design, analyze and optimize EM systems without the need for building physical prototypes, thus helping to shorten the development cycles and consequently cut costs.

The industry standard for simulating EM problems is using either the finite difference method or the finite element method (FEM). Optimization of the design process using such methods requires significant computational resources and time. With the emergence of artificial intelligence, along with specialized tools for automatic differentiation, the use of DL has become computationally much more efficient and cheaper. These advances in machine learning have ushered in a new era in EM simulations where engineers can compute results much faster while maintaining a certain level of accuracy.

This paper proposed two different models that can compute the magnetic field distribution in EM systems. The first model is based on a recurrent neural network, which is trained through a data-driven supervised learning method. The second model is an extension to the first with the incorporation of additional physics-based information to the authors’ model. Such a DL model, which is constrained by the laws of physics, is known as a physics-informed neural network. The solutions when compared with the ground truth, computed using FEM, show promising accuracy for the authors’ DL models while reducing the computation time and resources required, as compared to previous implementations in the literature.

The paper proposes a neural network architecture and is trained with two different learning methodologies, namely, supervised and physics-based. The working of the network along with the different learning methodologies is validated over several EM problems with varying levels of complexity. Furthermore, a comparative study is performed regarding performance accuracy and computational cost to establish the efficacy of different architectures and learning methodologies.

To discuss synthesis and evaluation of organo‐metallic chalcones as second‐order nonlinear optical (SONLO) materials.

The new chalcones have been synthesised via Knovoenagel reactions of ferrocen carboxaldehyde with two active methylene compounds.

The ferrocenyl chalcones prepared have shown bathochromic shift and thermal stability in polymeric film. On heating the dye films up to 80°C the extent of degradation reached up to 12 per cent and very small amount of degradation was observed at 43 and 60°C.

The paper shows that these compounds have UV‐Vis bathochromic shift, enabling them to be used as SONLO materials in the blue domain as well as dyes.

The representation of Heterogeneous Object (HO) is divided into two categories: Data model (DM) and material evaluation paradigm (MEP). A hybrid methodology with geometry model and volumetric dataset to represent heterogeneous properties is proposed in this paper. Geometry model of an object can guarantee the accuracy of the final HO slices; and volumetric dataset lends the flexible manipulability and other advantages to HO representation. Two MEPs, namely distance field (DF) based and Fixed Reference Features & Active Grading Source(s) (FRF&AGS) are presented to facilitate the process of HO representation according to the designer)s input parameters. The DM can be modified interactively with users until the final satisfactory result is obtained. In this paper, a scheme of HO slicing is described. In this method, we utilize the slices contour of geometrical model as constraint to reconstruct the HO slices, which can theoretically achieve the same accuracy with the geometrical shape. Some examples of Heterogeneous object represented with our scheme are provided.

To discuss synthesis and evaluation of organo‐metallic dyes as second‐order nonlinear optical (SONLO) material.

New dyes have been synthesised via Knovoenagel reactions of ferrocene carboxyaldehyde and two active methylene compounds.

The ferrocenyl dyes prepared have shown bathochromic shift and thermal stability.

These compounds have UV‐Vis bathochromic shift, enabling them to be used as SONLO materials as well as dyes.

The paper provides further information on the thermal studies of these types of molecules.

The purpose of this paper is to provide an efficient tool for simulating electrospinning process in virtual 3D space and optimizing experimental parameters. The fiber orientation from virtual or real electrospinning process can be easily measured using the image analysis technique. Using the semi-implicit Euler integration, the time integration can be more fast and stable, which enabled optimization of the electrospinning process. Also boundary conditions can be easily adopted during conjugate gradient matrix solving step.

To simulate the electrospinning process, the authors have adopted a particle-based modeling technique using the molecular dynamics theory, which is known to be suitable for modeling materials with nonlinear and nonhomogeneous behavior such as fibers or fabrics. Gravitational, tensional, and electrostatical forces and their Jacobians were carefully defined and chosen to maintain the stability of the governing equation. Preconditioned conjugate gradient method was used to solve the matrix iteratively with boundary conditions. The 2-D metaball fitting technique, which was applied in the previous research (Sul et al., 2009) on experimental nanofiber scanning electron microscopy images, was utilized with virtual nanofiber images. A staircase function and a new shading language were proposed to automatically calculate the orientation and radius distribution of the graphically simulated electrospun fiber structures. The automatic measurement procedure was verified via graphically designed virtual replica images. Also the orientation tendency acquired from the simulation was compared with that of experimental data.

Simulation result of fiber orientation showed linear relationship with the collecting drum speed. Use of particle-based method generated a simple system to simulate electrospinning process.

The semi-implicit Euler integration was applied to the electrospinning process and the final linear system was numerically stable to solve.

The purpose of this paper is to investigate the influence of geometrical microstructure of items obtained by applying a three-dimensional (3D) printing technology on their mechanical strength.

Three-dimensional printed items (3DPI) are composite structures of complex internal constitution. The buildup of the finite element (FE) computational models of 3DPI is based on a multi-scale approach. At the micro-scale, the FE models of representative volume elements corresponding to different additive layer heights and different thicknesses of extruded fibers are investigated to obtain the equivalent non-linear nominal stress–strain curves. The obtained results are used for the creation of macro-scale FE models, which enable to simulate the overall structural response of 3D printed samples subjected to tensile and bending loads.

The validation of the models was performed by comparing the computed results against the experimental ones, where satisfactory agreement has been demonstrated within a marked range of thicknesses of additive layers. Certain inadequacies between computed against experimental results were observed in cases of thinnest and thickest additive layers. The principle explanation of the reasons of inadequacies takes into account the poorer quality of mutual adhesion in case of very thin extruded fibers and too-early solidification effect.

Flexural and tensile experiments are simulated by FE models that are created with consideration to microstructure of 3D printed samples.

A method for the prediction of solder joint cycle life in surface‐mount assemblies is presented, based on the conversion of plastic solder shear strain into cycle life by means of an equation derived by Engelmaier. The paper introduces a different analytical procedure for the determination of solder joint shear strain. Shear strain is normally calculated from temperature and TCE differentials between package and interconnect board without consideration of elastic deformations. The suggested method derives average plastic shear strain of the solder joint at maximum temperature excursion from finite‐element analysis of a simple model consisting of an interconnect board, a solder joint and a package. All materials in the model have linear (elastic) properties, except solder which has non‐linear (elastic/plastic) characteristics. The solder stress/strain curve is described to the finite‐element programme with temperature‐dependent bilinear approximations. The solder joint is modelled as a single finite element so that only one value is computed for the plastic shear strain in the solder joint. This value represents the average shear strain which is converted into solder joint cycle life. The cycle life predictions with the finite‐element method are confirmed by cycling results obtained on actual hardware. The described method can serve as a design tool in the optimisation of surface‐mount assemblies. The procedure can help to define accelerated temperature cycling conditions.

This paper discusses the issues involved in the development of combined finite/discrete element methods; both from a fundamental theoretical viewpoint and some related algorithmic considerations essential for the efficient numerical solution of large scale industrial problems. The finite element representation of the solid region is combined with progressive fracturing, which leads to the formation of discrete elements, which may be composed of one or more deformable finite elements. The applicability of the approach is demonstrated by the solution of a range of examples relevant to various industrial sections.

The optimal material microstructures in pure material design are no longer efficient or optimal when accounting macroscopic structure performance with specific boundary conditions. Therefore, it is important to provide a novel multiscale topology optimization framework to tailor the topology of structure and the material to achieve specific applications. In comparison with porous materials, composites consisting of two or more phase materials are more attractive and advantageous from the perspective of engineering application. This paper aims to provide a novel concurrent topological design of structures and microscopic materials for thermal conductivity involving multi-material topology optimization (material distribution) at the lower scale.

In this work, the effective thermal conductivity properties of microscopic three or more phase materials are obtained via homogenization theory, which serves as a bridge of the macrostructure and the periodic material microstructures. The optimization problem, including the topological design of macrostructures and inverse homogenization of microscopic materials, are solved by bi-directional evolutionary structure optimization method.

As a result, the presented framework shows high stability during the optimization process and requires little iterations for convergence. A number of interesting and valid macrostructures and material microstructures are obtained in terms of optimal thermal conductive path, which verify the effectiveness of the proposed mutliscale topology optimization method. Numerical examples adequately consider effects of initial guesses of the representative unit cell and of the volume constraints of adopted base materials at the microscopic scale on the final design. The resultant structures at both the scales with clear and distinctive boundary between different phases, making the manufacturing straightforward.

This paper presents a novel multiscale concurrent topology optimization method for structures and the underlying multi-phase materials for thermal conductivity. The authors have carried out the concurrent multi-phase topology optimization for both 2D and 3D cases, which makes this work distinguished from existing references. In addition, some interesting and efficient multi-phase material microstructures and macrostructures have been obtained in terms of optimal thermal conductive path.

Analyses were performed during the conceptual design stage of a 20in. threaded connector for deep water J‐pipe lay, as part of a research project developed by Tecnomare and partly funded by the EEC. The joint consists of two parts, namely a pin and a box, provided with cylindrical threads. It was essential for the joint design to be fully leak‐proof for both internal and external pressure and this requirement had to be satisfied also under the maximum bending moment allowable for the sealine. Sealing was accomplished on a cone surface by screwing the pin into the box until yield was reached. The FEM analysis was carried out primarily to check that the pin and box remain pressed to one another over the sealing surface in every design condition with adequate pressure to prevent leakage. For this purpose, the analysis was a powerful design technique, as it gave an easy understanding of the structural behaviour and provided proper stiffness by making the joint either larger or thinner wherever required. The main characteristic of this work is that FEM analysis has been utilized as a design method rather than as a check. The analysis was performed by means of ADINA (Automatic Dynamic Incremental Non‐linear Analysis) program. Contact pressure between sealing surfaces, as achieved during the joint screwing phase, was modelled through thermal elongation. Pressure loads and external forces were superimposed through a step‐by‐step procedure, by accounting for the elastoplastic behaviour all around the sealing surface. In order to verify the behaviour of the mechanical joint, six prototypes have been fabricated and tested under the design loads of the lay phase and the operative life. The results of the tests confirmed the correct design and the results of non‐linear finite element analysis. The most important performances of the joint can be summarized as follows: (1) the make‐up phase is rapid and easy: no problems of frictional pick‐up took place; (2) no leakage happened during the internal pressure tests: the pressure of 300atm (1.5 times the design internal pressure) was maintained for 12h; (3) the load conditions of the second series of tests were: 200atm of internal pressure and the maximum allowable bending moment relevant to the pipe: after 2h no leakage happened. This paper describes the model used for the analysis, discusses its implications and the most important results achieved in comparison with the tests of the experimental phase.