Search

Aluminium metal production through electrolytic reduction of alumina in a cryolite bath is a complex, multi-physics, multi-scale process, including magneto-hydrodynamics (MHD), bubble flow, thermal convection, melting and solidification phenomena based on a set of chemical reactions. Through interactions of the different forces applied to the liquid bath combined with the different time and length scales, self-organised fluctuations occur. Moreover, the MHD behaviour causes a complex metal pad profile and a series of surface waves due to the meta-stable condition of the metal/cryolite interface. The large aspect ratio of an industrial cell, with a footprint of 20 by 4 m and at the same time having dimensions approaching just 30 mm of height for the reaction zone, prevents an integrated approach where all relevant physics are included in a single mathematical model of this large degree of freedom system. In order to overcome these challenges, different modelling approaches have been established in ANSYS ® FLUENT ® ; Three models are used to predict details of specific physics: one to predict the electro-magnetic forces and hence the metal pad profile, a second that resolves details of the local bubble dynamics around a single anode and a third for the full cell bath flow. Results from these models are coupled to allow integration of the different phenomena into a full cell alumina distribution model. The current paper outlines each of the approaches and presents how the coupling between them can be realized in a complete framework, aiming to provide new insight into the process. [ABSTRACT FROM AUTHOR]

In this paper we report the extension of an earlier reported DNS method (Deen et al., 2012; Deen and Kuipers, 2013) based on a novel Immersed Boundary Method (IBM) which incorporates the fluid-solid coupling at the level of the discrete field equations. The extended method is used to study coupled mass and heat transport in dense fluid-particle systems where the coupling arises as a consequence of an exothermal chemical reaction proceeding at the exterior surface of the particles. Following a detailed verification (using an independent numerical technique) and validation (using established empirical correlations) we apply our DNS technique to study coupled mass and heat transfer in a dense fluid-particle system. In addition a comparison is made with results obtained from a simple one-dimensional (1D) heterogeneous reactor model which uses empirical closures for the fluid-particle mass and heat transfer coefficients. The main features of the complex transient temperature profiles obtained from our DNS agree quite well with the corresponding profiles obtained from the 1D heterogeneous reactor model. [ABSTRACT FROM AUTHOR]

In-plane centroids of textile composites are simulated as cross-correlated random fields. Each tow position is defined as an average trend quantified from experimental data, added with zero-mean deviations produced as a stochastic field. Realisations of these fields are generated using a framework based on the Karhunen–Loève series expansion that is calibrated with experimental information from prior work. Positional deviations are obtained that are correlated along the tow and between neighbouring tows. The application is a 2/2 twill woven carbon fibre reinforced epoxy consisting of multiple unit cells. Generated in-plane deviations of the warp and weft tows resemble the experimental fluctuations with similar wavelengths. Simulation of thousand specimens demonstrates that the virtual in-plane positions possess the experimental standard deviation and correlation lengths on average. [ABSTRACT FROM AUTHOR]

Abstract: Pore-scale simulations of transport phenomena in foam based catalytic converters are highly desirable due to their intrinsic accuracy in describing the involved physics. Unfortunately, such computations are prohibitively time consuming and costly. Here we combine a detailed pore-scale characterization with a volume average description of an open-cell foam based catalytic converter. The resulting volume averaged model is accurate and more than three orders of magnitude more efficient in terms of computational time, compared to direct pore scale simulations. This makes simulations of the related complex phenomena tractable, providing a highly flexible and powerful tool for parametric studies of open-cell foams. The volume-averaged model is validated with a pore-scale simulation of a full-scale catalytic converter used for automotive exhaust gas treatment. Fast chemistry is assumed and the heat released by the chemical reaction in combination with conjugate heat transfer is considered. We show that the conversion to pressure drop trade-off improves with increasing foam porosity and the heat release of the reaction significantly affects the pressure drop across the reactor. The simulations show non-uniform surface temperatures which are attributed to the non-unity Lewis number of the mass transfer limiting species. The model predictions agree well with experimental measurements of pressure drop and Sherwood number in open-cell foams. [Copyright &y& Elsevier]

The understanding and the prediction of the failure behaviour of materials with pronounced microstructural effects is of crucial importance. This paper presents a novel computational methodology for the handling of fracture on the basis of the microscale behaviour. The basic principles presented here allow the incorporation of an adaptive discretization scheme of the structure as a function of the evolution of strain localization in the underlying microstructure. The proposed quasi-discrete methodology bridges two scales: the scale of the material microstructure, modelled with a continuum type description; and the structural scale, where a discrete description of the material is adopted. The damaging material at the structural scale is divided into unit volumes, called cells, which are represented as a discrete network of points. The scale transition is inspired by computational homogenization techniques; however it does not rely on classical averaging theorems. The structural discrete equilibrium problem is formulated in terms of the underlying fine scale computations. Particular boundary conditions are developed on the scale of the material microstructure to address damage localization problems. The performance of this quasi-discrete method with the enhanced boundary conditions is assessed using different computational test cases. The predictions of the quasi-discrete scheme agree well with reference solutions obtained through direct numerical simulations, both in terms of crack patterns and load versus displacement responses. [ABSTRACT FROM AUTHOR]

An unconventional multi-scale approach is investigated to model interfacial damage and subsequent effects in highly filled particulate composites such as solid propellants. The basic framework by Nadot et al. (2006 Damage modelling framework for viscoelastic particulate composites via a scale transition approach, Journal of Theoretical and Applied Mechanics 44:553–583) is enhanced to deal with damage state and configuration evolution under any loading. Three criteria are formulated at the microscale. The first one concerns the nucleation of new defects, while the other two (for closure and re-opening) monitor the respective proportions of open and closed defects. The formulation exploits the explicit microstructure representation and the knowledge of the local displacement field as a function of local morphology around the interfaces. The multi-scale model is finally implemented through a numerical solving procedure able to deal with successive and/or simultaneous discrete damage events (nucleation/closure/re-opening of defects) depending on the loading path. Numerical illustrations are given for a three-dimensional microstructure containing 351 particles submitted to complex loading sequence involving extension, shear and contraction stages. The results show the ability of the model to estimate damage characteristics (position, orientation within the material) in addition to induced effects at both microscale and macroscale (induced anisotropy, unilateral effects and influence on local fields). [ABSTRACT FROM PUBLISHER]

Abstract: A multi-scale modelling approach is employed to investigate the elasto-plastic behaviour of the pure chitosan biological polymer, as well as its composite when it is embedded with a graphene sheet, and a carbon nanotube. The model directly incorporates inter-atomic potentials, describing the energetics of molecular clusters, into a continuum-based computation without any need for a parameter fitting. The coupling of the atomistic and continuum levels is achieved via the adoption of the Cauchy–Born rule which provides a suitable framework for investigating the mechanical behaviour of materials undergoing large deformations. All data on atomic bonding, bond angle, bond torsion, and nonbonding interactions relevant to the molecular clusters are incorporated into the constitutive continuum model. The continuum point within a finite element (FE) mesh is represented by an atomic cluster, referred to as a nanoscopic representative volume element (NRVE). We have introduced a parameter R that represents the ratio of the volume of finite element within which the NRVE is embedded to the volume of the NRVE. It is shown that the variation of R directly affects the results on the stress–strain behaviour of the composite, and that for R ∼1.0, our computed results on the elastic modulus of the simple and composite systems are fair in agreement with the experimental data. [Copyright &y& Elsevier]