E12
Multiscale Simulation

Back to overview

13:20
conference time (CEST, Berlin)
Computing FLD Diagrams and Effective Properties of Polycrystalline Metals
27/10/2021 13:20 conference time (CEST, Berlin)
Room: E
G. Lemoine (Hexagon, LUX)
G. Lemoine (Hexagon, LUX)
Polycrystalline metallic materials are essential components in modern automotive and aircraft industries (e.g., high-performance components of fuel cell systems, gears, engines…). The mechanical properties of these components and materials are significantly influenced by their structure at different scales. Enabling the microstructure to become an integral part of the design process of metal components is a key challenge to enhance the usage of new advanced metal alloys for high-performance and light-weighting applications. Computational Homogenization of Polycrystals (CHP) arises as an ideal tool to perform virtual material engineering, providing a relation between material microstructure and its performance. CHP is based on the simulation (either using Finite Elements FE or Fast Fourier Transform based solvers) of the mechanical response of Representative Volume Elements (RVE) of the polycrystalline microstructure under imposed loading paths and environmental conditions. Digimat-MF and Digimat-FE propose a complete framework to conduct such simulations. This framework includes polycrystal and texture generation, crystal plasticity, mean-field approach and FE/FFT solvers. The presentation will illustrate the efficiency of that workflow to predict texture evolution in forming processes, strength, and evolution of plastic anisotropy. Such simulation workflow can also be applied to perform a localization study. Starting from a cold forming simulation within Simufact, the loading path seen by a particular material point in the forming FE model is applied to the polycrystalline RVE to determine the local evolution of the texture and the yield locus. Finally, another aspect of Digimat-MF devoted to forming-limit diagrams (FLD) will be presented. Sheet formability is commonly evaluated based on such FLD in the sheet-metal forming industry. However, the experimental measurement of FLD is a difficult, time consuming and expensive process. Exporting formability of metallic sheets from our virtual testing tools replaces many of the experimental measurements. In this presentation, we will analyze forming-limit strains of metals using J2 plasticity or crystal plasticity mean-field models in conjunction with an imperfection-based approach.
FLD Diagrams, Polycrystalline metals, metals, microstructure, FFT,Computational homogenization
13:40
conference time (CEST, Berlin)
Predictive Modeling of Void Closure During the Hot Rolling of Bars Using Finite Element Analysis
27/10/2021 13:40 conference time (CEST, Berlin)
Room: E
C. Pondaven, B. Erzar (ABS Centre Métallurgique, FRA); J.M. Colomer (Datadvance, FRA)
C. Pondaven, B. Erzar (ABS Centre Métallurgique, FRA); J.M. Colomer (Datadvance, FRA)
The hot rolling of bars issued from continuous casting is a manufacturing process aiming to achieve the desired product cross-section, the refinement of material structure and the annihilation of internal defects such as shrinkage porosity. The mastering of the void closure phenomenon during rolling is crucial to guarantee the internal soundness of finished products. The pore closure process is strongly influenced by the defects dimensions and orientation related to the applied loadings which tends to limit the application of classical models to estimate void closure during complex forming route. In this study, a methodology is developed to predict the evolution of a single pore undergoing a complex loading route thanks to finite element computations, design of experiments, and surrogate modeling techniques. The mechanical loading during the first stand of an industrial rolling route is firstly characterized by measurements performed at the center of the product using numerical modeling. The loadings are secondly applied as boundary conditions of a simulation performed on a Representative Volume Element (RVE) containing an ellipsoid defect. This scale transition is carried out to reduce the computational time needed to evaluate the void evolution using an explicit representation of the cavity. A design of experiments (DoE) is then generated by means of the pSeven® software and consists in defining several combinations of the values of the geometrical parameters of the void (main axes dimensions and three rotation angles) integrated in the RVE. For each morphology, the automatic post-processing of the numerical results allows computing the void evolution as the ratio between the final pore volume and its initial volume. Eventually, the dataset is used to build a surrogate model. Knowing the initial geometry of the void, this model provides an almost immediate estimation of the final volume of a pore undergoing a loading route representative of hot rolling.
Void closure, design of experiments, finite element method, hot rolling, surrogate modeling
14:00
conference time (CEST, Berlin)
Multi-scale Modelling to Estimate Wafer Bow in 3D NAND Applications
27/10/2021 14:00 conference time (CEST, Berlin)
Room: E
S. Varadharajan, R. Patil (Lam Research India Pvt. Ltd., IND)
S. Varadharajan, R. Patil (Lam Research India Pvt. Ltd., IND)
Flash memory is widely used in computers, mobile devices and other electronic gadgets with the NAND logic gate based memory architecture commonly found in USB drives, memory cards and solid-state drives. NAND memory architecture offers higher storage density compared other memory architectures and were originally built as a 2D planar structure over the silicon wafer. To further improve the memory density, these 2D NAND structures were stacked on top of one another to create a complex 3D memory architecture. 3D NAND is the state of the art in flash memory technology and the number of stack layers exceeds 128 in many applications. Intro- ducing a higher number of film layers results in significant bowing of the silicon wafer substrate due to mismatch of thermal expansion coefficients between the different film layers within the stack combined with anisotropy introduced due to directional layout of devices on the wafer. Understanding and estimating the bowing of the substrate is crucial to successful execution of subsequent wafer processing steps like etching, deposition and lithography. A multi-scale modelling technique using a commercial finite element method (FEM) package is presented to estimate the wafer bow for 3D NAND applications. 3D NAND device structures on the wafer are periodic in nature at the micro-level. Three-dimensional representative volume element (RVE) method based on numerical homogenisation technique is used to estimate the effective material properties such as elastic constants and thermal expansion coefficients of these structures. The computed effective properties at the micro-level are mapped on to the die level that are at the meso-scale. The meso-scale consists of periodically patterned individual dies on the wafer with the necessary circuitries along with the 3D NAND devices. Two-dimensional RVE is employed at the meso-scale to estimate the effective elastic constants and thermal expansion coefficients of the dies. The macro bowing behaviour of the wafer is evaluated using the effective proper- ties computed at the micro and the meso scales. The bowing of the wafer along the bit-line and word-line directions are evaluated on cooling the wafer from a given process temperature to room temperature. Sensitivity studies are performed to understand the influence of various geometric and material properties of the given NAND structure on the overall macro behaviour of the wafer.
Multi-scale Modelling, Representative Volume Element, Numerical Homogenisation, Finite Element Method, Wafer Bow
14:20
conference time (CEST, Berlin)
Fast Prediction of the Effective Properties of 3D Woven Composites
27/10/2021 14:20 conference time (CEST, Berlin)
Room: E
A. Cheruet (Hexagon, LUX); G. Lemoine (Hexagon, BEL)
A. Cheruet (Hexagon, LUX); G. Lemoine (Hexagon, BEL)
Three dimensional woven architectures of continuous fibres give the possibility to add a 3rd direction of reinforcement with the addition of yarns interlacing through the thickness. The additional benefit of this interlacing yarn through the thickness is to improve damage tolerance properties. However, this opportunity of adding a 3rd dimension to the weave pattern creates an extremely wide variety of possible combination of interlacement which renders classical methods of screening based on physical prototyping extremely costly and time consuming for material developers and engineers. To support the development of 3D woven composites and to provide the industry with a quick screening of potential weave pattern properties that meet the mechanical specifications, numerical simulation tools must be proposed to designers to analyze and predict stiffness, failure and thermo-mechanical behavior of any 3D woven representative volume element (RVE). A complete multi-scale workflow is proposed, starting from the constituents’ properties to quickly provide the effective properties of each weave pattern under consideration. In addition to the workflow proposes, the possibility to generate an ideal geometry of the weave pattern or to import a realistic weave pattern describing better the manufacturing steps will propose a more accurate variation of the yarn properties. Variable properties of the yarn are computed using the local fiber volume content in the yarn cross-section. The proposed approach will be able to allow for elasticity and plasticity with failure in both contexts. Finally, the weave pattern problem can be solved, either using a FFT solver (Fast Fourier Transformation solver) allowing a generation of the engineering constants in an extremely efficient and fast process or either using the finite-element method which is able to accomodate non-linearity with failure to be predicted accurately. In the presentation the efficiency of the proposed workflow will be illustrated by applying it on real weave patterns.
3D wovens, CFRP, continuous fibre, FFT, interlacing
×

[TITLE]

[LISTING

[ABSTRACT]

[DATE]

[ROOM]

[KEYWORDS]