Three parallel algorithms for classical molecular dynamics are presented. The first assigns each processor a fixed subset of atoms; the second assigns each a fixed subset of inter-atomic forces to compute; the third assigns each a fixed spatial region. The algorithms are suitable for molecular dynamics models which can be difficult to parallelize efficiently—those with short-range forces where the neighbors of each atom change rapidly. They can be implemented on any distributed-memory parallel machine which allows for message-passing of data between independently executing processors. The algorithms are tested on a standard Lennard-Jones benchmark problem for system sizes ranging from 500 to 100,000,000 atoms on several parallel supercomputers--the nCUBE 2, Intel iPSC/860 and Paragon, and Cray T3D. Comparing the results to the fastest reported vectorized Cray Y-MP and C90 algorithm shows that the current generation of parallel machines is competitive with conventional vector supercomputers even for small problems. For large problems, the spatial algorithm achieves parallel efficiencies of 90% and a 1840-node Intel Paragon performs up to 165 faster than a single Cray C9O processor. Trade-offs between the three algorithms and guidelines for adapting them to more complex molecular dynamics simulations are also discussed.

Fast parallel algorithms for short-range molecular dynamics

Masonry structures, although classically suitable to withstand gravitational loads, are sensibly vulnerable if subjected to extraordinary actions such as earthquakes, exhibiting cracks even for events of moderate intensity compared to other structural typologies like as reinforced concrete or steel buildings. In the last half-century, the scientific community devoted a consistent effort to the computational analysis of masonry structures in order to develop tools for the prediction (and the assessment) of their structural behavior. Given the complexity of the mechanics of masonry, different approaches and scales of representation of the mechanical behavior of masonry, as well as different strategies of analysis, have been proposed. In this paper, a comprehensive review of the existing modeling strategies for masonry structures, as well as a novel classification of these strategies are presented. Although a fully coherent collocation of all the modeling approaches is substantially impossible due to the peculiar features of each solution proposed, this classification attempts to make some order on the wide scientific production on this field. The modeling strategies are herein classified into four main categories: block-based models, continuum models, geometry-based models, and macroelement models. Each category is comprehensively reviewed. The future challenges of computational analysis of masonry structures are also discussed.

/pdf/modeling-strategies-for-the-computational-analysis-of-4lh0cu9a67.pdf

Modeling Strategies for the Computational Analysis of Unreinforced Masonry Structures: Review and Classification

A 3D detailed micro-model for the in-plane and out-of-plane numerical analysis of masonry panels

http://repositorium.sdum.uminho.pt/bitstream/1822/68377/1/ManuscriptStructures.pdf

Simulation of the in-plane structural behavior of unreinforced masonry walls and buildings using DEM

The importance of crack propagation in the structural behaviour of concrete and masonry structures has led to the development of a wide range of finite element methods for crack simulation. A common standpoint in many of them is the use of tracking algorithms, which identify and designate the location of cracks within the analysed structure. In this way, the crack modelling techniques, smeared or discrete, are applied only to a restricted part of the discretized domain. This paper presents a review of finite element approaches to cracking focusing on the development and use of tracking algorithms. These are presented in four categories according to the information necessary for the definition and storage of the crack-path. In addition to that, the most utilised criteria for the selection of the crack propagation direction are summarized. The various algorithmic issues involved in the development of a tracking algorithm are discussed through the presentation of a local tracking algorithm based on the smeared crack approach. Challenges such as the modelling of arbitrary and multiple cracks propagating towards more than one direction, as well as multi-directional and intersecting cracking, are detailed. The presented numerical model is applied to the analysis of small- and large-scale masonry and concrete structures under monotonic and cyclic loading.

/pdf/challenges-tools-and-applications-of-tracking-algorithms-in-4455wxlwt6.pdf

Challenges, Tools and Applications of Tracking Algorithms in the Numerical Modelling of Cracks in Concrete and Masonry Structures

Micro-scale continuous and discrete numerical models for nonlinear analysis of masonry shear walls

Multiscale thermo-mechanical analysis of multi-layered coatings in solar thermal applications

Adaptive and off-line techniques for non-linear multiscale analysis

The increased complexity of the materials used for thermo-mechanical applications implies the need of taking into account the microstructure and assessing its behavior. This paper proposes an extension of classical first-order computational homogenization to involve thermo-mechanical effects. The formulation used considers as fully coupled the two physics of the problem, firstly determining the thermal parameters and then considering thermal effects onto the mechanical behavior. The work is completed by benchmark tests to assess the behavior of the new model.

Thermo-mechanical homogenization of composite materials

In recent years, the study of the behavior of materials at the microscopic level has increased significantly in terms of design of high-performance materials. Despite recent advances in high-performance computers, the application of multiscale numerical methods to simulate large structures still requires prohibitive computational costs. The purpose of this work is to provide a procedure capable to predict the nonlinear mechanical response of composite materials under monotonic incremental load to reduce the computational cost required for the numerical analysis of complex structures. The solution of the macroscopic problem through the first order multiscale method (FE2) will be replaced by a Discrete Multiscale Model (DM) characteristic of the Representative Volume Element (RVE). Through the definitions of an equivalent damage parameter (deq), function of the global stress at the microscale, a series of virtual tests in deformation control will be carried out, storing the stress-strain state reached by certain levels of deq in a Database. Analyzing the evolution of the fracture in the composite materials can be observed as the non-linear regime is reached only in some elements of the structure. Therefore, a procedure, the Discrete Multiscale Threshold Surface (DMTS), in which the RVE analysis is used to obtain the surface at which the damage begins (deq>0) is proposed. This law allows to know if for a certain stress-strain state the material has damaged, without needing to solve the micro-model. Once the damage is initiated, it is proposed to generate an RVE only at the integration points that have been damaged. This work will demonstrate how the FE2 method can be replaced by a Discrete Multiscale Model, representative of the composite material, obtaining significant computational improvements.

https://revista.aemac.org/materiales-compuestos/article/download/174/119

Stefano Zaghi

Papers

Micro-scale continuous and discrete numerical models for nonlinear analysis of masonry shear walls

Multiscale thermo-mechanical analysis of multi-layered coatings in solar thermal applications

Adaptive and off-line techniques for non-linear multiscale analysis

Thermo-mechanical homogenization of composite materials

Nuevo Modelo Discreto Multiescala (DM) para análisis no-lineales de materiales compuestos