scholarly journals One for all: Universal material model based on minimal state-space neural networks

2021 ◽  
Vol 7 (26) ◽  
pp. eabf3658
Author(s):  
Colin Bonatti ◽  
Dirk Mohr

Computational models describing the mechanical behavior of materials are indispensable when optimizing the stiffness and strength of structures. The use of state-of-the-art models is often limited in engineering practice due to their mathematical complexity, with each material class requiring its own distinct formulation. Here, we develop a recurrent neural network framework for material modeling by introducing “Minimal State Cells.” The framework is successfully applied to datasets representing four distinct classes of materials. It reproduces the three-dimensional stress-strain responses for arbitrary loading paths accurately and replicates the state space of conventional models. The final result is a universal model that is flexible enough to capture the mechanical behavior of any engineering material while providing an interpretable representation of their state.

Author(s):  
Marzieh Azarnoosh ◽  
Marcus Stoffel ◽  
Dieter Weichert

Over the last several decades investigations of replacement material for intervertebral disc (IVD) have been an important topic in medical research. The challenge is to create materials whose mechanical behavior ideally matches that of the articular cartilage comprising the native discs. Thus, the study of articular cartilages underlying mechanical characteristics is a key issue for the successful development and refinement of replacement materials. Using both experimental and cartilage histostructural data, including fiber orientation, a visco-hyperelastic-diffusion (VHD) material model is developed and implemented. This allows us to numerically study the mechanical behavior of an IVD consisting of a cartilaginous ring surrounding a fluid core. In this work, a three dimensional finite element (FE) model is developed to simulate the behavior of an IVD under various loading conditions. Finally, model parameters are iteratively determined by comparing the simulation results to compression tests on corresponding discs performed in a MTS machine with a tempered nutrient solution.


Author(s):  
D. Liu ◽  
S. Koric ◽  
A. Kontsos

Direct numerical simulations (DNS) of knitted textile mechanical behavior are for the first time conducted on high performance computing (HPC) using both the explicit and implicit finite element analysis (FEA) to directly assess effective ways to model the behavior of such complex material systems. Yarn-level models including interyarn interactions are used as a benchmark computational problem to enable direct comparison in terms of computational efficiency between explicit and implicit methods. The need for such comparison stems from both a significant increase in the degrees-of-freedom (DOFs) with increasing size of the computational models considered as well as from memory and numerical stability issues due to the highly complex three-dimensional (3D) mechanical behavior of such 3D architectured materials. Mesh and size dependency, as well as parallelization in an HPC environment are investigated. The results demonstrate a satisfying accuracy combined with higher computational efficiency and much less memory requirements for the explicit method, which could be leveraged in modeling and design of such novel materials.


Materials ◽  
2021 ◽  
Vol 14 (3) ◽  
pp. 691
Author(s):  
Francisco-José Gallardo-Basile ◽  
Yannick Naunheim ◽  
Franz Roters ◽  
Martin Diehl

Lath martensite is a complex hierarchical compound structure that forms during rapid cooling of carbon steels from the austenitic phase. At the smallest, i.e., ‘single crystal’ scale, individual, elongated domains, form the elemental microstructural building blocks: the name-giving laths. Several laths of nearly identical crystallographic orientation are grouped together to blocks, in which–depending on the exact material characteristics–clearly distinguishable subblocks might be observed. Several blocks with the same habit plane together form a packet of which typically three to four together finally make up the former parent austenitic grain. Here, a fully parametrized approach is presented which converts an austenitic polycrystal representation into martensitic microstructures incorporating all these details. Two-dimensional (2D) and three-dimensional (3D) Representative Volume Elements (RVEs) are generated based on prior austenite microstructure reconstructed from a 2D experimental martensitic microstructure. The RVEs are used for high-resolution crystal plasticity simulations with a fast spectral method-based solver and a phenomenological constitutive description. The comparison of the results obtained from the 2D experimental microstructure and the 2D RVEs reveals a high quantitative agreement. The stress and strain distributions and their characteristics change significantly if 3D microstructures are used. Further simulations are conducted to systematically investigate the influence of microstructural parameters, such as lath aspect ratio, lath volume, subblock thickness, orientation scatter, and prior austenitic grain shape on the global and local mechanical behavior. These microstructural features happen to change the local mechanical behavior, whereas the average stress–strain response is not significantly altered. Correlations between the microstructure and the plastic behavior are established.


Materials ◽  
2021 ◽  
Vol 14 (15) ◽  
pp. 4092
Author(s):  
Kamil Bacharz ◽  
Barbara Goszczyńska

The paper reports the results of a comparative analysis of the experimental shear capacity obtained from the tests of reinforced concrete beams with various static schemes, loading modes and programs, and the shear capacity calculated using selected models. Single-span and two-span reinforced concrete beams under monotonic and cyclic loads were considered in the analysis. The computational models were selected based on their application to engineering practice, i.e., the approaches implemented in the European and US provisions. Due to the changing strength characteristics of concrete, the analysis was also focused on concrete contribution in the shear capacity of reinforced concrete beams in the cracked phase and on the angle of inclination of diagonal struts. During the laboratory tests, a modern ARAMIS digital image correlation (DIC) system was used for tracking the formation and development of diagonal cracks.


2007 ◽  
Vol 345-346 ◽  
pp. 1469-1472
Author(s):  
Gab Chul Jang ◽  
Kyong Ho Chang ◽  
Chin Hyung Lee

During manufacturing the welded joint of steel structures, residual stress is produced and weld metal is used inevitably. And residual stress and weld metal influence on the static and dynamic mechanical behavior of steel structures. Therefore, to predict the mechanical behavior of steel pile with a welded joint during static and dynamic deformation, the research on the influence of the welded joints on the static and dynamic behavior of steel pile is clarified. In this paper, the residual stress distribution in a welded joint of steel piles was investigated by using three-dimensional welding analysis. The static and dynamic mechanical behavior of steel piles with a welded joint is investigated by three-dimensional elastic-plastic finite element analysis using a proposed dynamic hysteresis model. Numerical analyses of the steel pile with a welded joint were compared to that without a welded joint with respect to load carrying capacity and residual stress distribution. The influence of the welded joint on the mechanical behavior of steel piles during static and dynamic deformation was clarified by comparing analytical results


2020 ◽  
Vol 29 (1) ◽  
pp. 1-8
Author(s):  
Ahmed Allali ◽  
Sadia Belbachir ◽  
Ahmed Alami ◽  
Belhadj Boucham ◽  
Abdelkader Lousdad

AbstractThe objective of this work lies in the three-dimensional study of the thermo mechanical behavior of a blade of a centrifugal compressor. Numerical modeling is performed on the computational code "ABAQUS" based on the finite element method. The aim is to study the impact of the change of types of blades, which are defined as a function of wheel output angle β2, on the stress fields and displacements coupled with the variation of the temperature.This coupling defines in a realistic way the thermo mechanical behavior of the blade where one can note the important concentrations of stresses and displacements in the different zones of its complex form as well as the effects at the edges. It will then be possible to prevent damage and cracks in the blades of the centrifugal compressor leading to its failure which can be caused by the thermal or mechanical fatigue of the material with which the wheel is manufactured.


Sign in / Sign up

Export Citation Format

Share Document