1,720,993 research outputs found
Crack growth propagation using standard FEM
No full textThis paper is part of research lines providing numerical methods for structural capability related to an evolving crack length in two-dimensional linear-elastic fracture mechanics of isotropic materials. We propose a numerical approach which efficiently simplifies more sophisticated approaches and yet, yield accurate results. Widely employed in academia, they are hardly used commercially unless ad hoc libraries are implemented. The proposed strategy improves early works where crack growth can be treated as a shape perturbation of the domain with existing cracks. The numerical results prove the accuracy of our strategy in comparison with both numerical and experimental outcomes present in literature
Physical design for distributed RFID-based supply chain management
No full textIn consumer products market, supply chain management (SCM) is a complex and significant issue in the governance of organizations, people with their activities, technology, information and resources involved in transferring a product or service from a supplier to a final customer. To this aim, radio-frequency identification (RFID) is a promising wireless technology allowing to link an object with its “virtual counterpart”, i.e., its representation within information systems. In this context, a SCM system has to face a huge amount of RFID data, generated in the tracking of supply chain resources. In particular when RFID installations become larger and more physically distributed, efficient and scalable analysis of such data becomes a concern. Currently, state of the art approaches provide hard-coded solutions where the processing of RFID data occurs in a central location; as the amount and distribution of data grow, the workload requires significant consumption of resources, and quickly outpaces the capacity of a centralized processing server. In this paper, we consider the problem of distributing the RFID processing workload—possibly huge—proposing the physical design of a scalable and distributed system. Such system is built on top of a general framework for SCM, based on the first principles of linear algebra, in particular, on tensorial calculus. We consider challenges in instantiating such a system in large distributed settings, and design techniques for distributed real time query processing. Experimental results, using large traces, demonstrate the efficiency and scalability of our proposal with respect to competing approaches
High-performance data structures for de novo assembly of genomes: cache oblivious generic programming
No full textReconstructing genomes of organisms from high-throughput sequencing experiments without a reference genome available (de novo assembly) is a challenging problem which has been approached in several ways in the past decade. Although numerous methods are available and many offer fair performance in reconstruction, there is a lack of generalized template libraries and interchangeable data structures/methods for serial, multithreaded and distributed processing. In this work we propose a novel set of cache oblivious generic data structures for serial, multithreaded and distributed processing of high-throughput sequencing data for the creation of de Bruijn or k-mer graphs towards their usage in de novo assembly and related HTS data analytics problems
A computational platform for carbon nanotube nanocomposites optimization
No full textA nonlinear 3D constitutive theory is implemented to describe the hysteresis exhibited by carbon nanotube nanocomposites due to the shear stick-slip between the carbon nanotubes and the polymer chains of the hosting matrix. The meso-scale theory is a combination of the Eshelby equivalent inclusion theory, the Mori-Tanaka homogenization method, and the introduction of inhomogeneous inclusions with inelastic eigenstrains besides the standard elastic eigenstrains. The shear stick-slip is accounted for as an incremental hysteretic eigenstrain whose evolution is regulated by the von Mises function of the deviatoric part of the interfacial stress discontinuity. The 3D material model is implemented in explicit dynamic form in a finite element platform called Fenics which makes use of a time integration scheme based on the Extended Average Mean Value Theorem and a special form of the Impulse-Momentum Law. The code written in Python and C++ has a layer structure and is fully optimized for fast computations. The code is embedded in a genetic-type optimization algorithm (Differential Evolutionary) to optimize the damping capacity exhibited by carbon nanotube nanocomposites made of thermosetting and thermoplastic polymers
Computational efficiency and accuracy of sequential nonlinear cyclic analysis of carbon nanotube nanocomposites
No full textThe accuracy and efficiency of a numerical strategy for sequential nonlinear cyclic analyses of carbon nanotube nanocomposites are investigated. The computational approach resorts to a nonlinear 3D finite element implementation that seeks to solve the cyclic hysteretic response of the nanocomposite. A variant of the Newton-Raphson method within a time integration scheme is proposed whereby the elastic tangent matrix is chosen as iteration matrix without paying the price of its iterative update. This is especially rewarding in the context of the employed mechanical model which exhibits hysteresis manifested through a discontinuous change in the stiffness at the reversal points where the loading direction is reversed. Key implementation aspects – such as the integration of the nonlinear 3D equations of motion, the numerical accuracy/efficiency as a function of the time step or the mesh size – are discussed. In particular, efficiency is regarded as performing fast computations especially when the number of cyclic analyses becomes large. By making use of laptop CPU cores, a good speed of computations is achieved not only through parallelization but also employing a caching procedure for the iteration matrix
- …
