Image-Based Random Fields in Numerical Modeling of Civil Engineering Problems

Result code in IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F68407700%3A21110%2F23%3A00382878" target="_blank" >RIV/68407700:21110/23:00382878 - isvavai.cz</a>

Result on the web

—

DOI - Digital Object Identifier

—

Result language

angličtina

Original language name

Image-Based Random Fields in Numerical Modeling of Civil Engineering Problems

Original language description

This thesis is focused on testing the random fields derived from image analysis in real engineering problems. In particular, we study two-dimensional stationary heat conduction and nonlinear damage mechanics problems. Our goal is to represent material heterogeneity and phase change using well-designed random field parameters. In this work, we utilize four structural patterns of materials; three are artificially generated with known parameters, while the last pattern is a scan of real material - the Liapor concrete. A subsequent study is performed on a total of six different versions of random fields constructed using Karhunen-Loève expansion. The random fields are distinguished by the different covariance kernels or by the modification of their continuous values to binary ones. The random fields are statistically compared with reference cut-outs from the original images, both in terms of the inputs to the material models and their model responses - temperature, displacement, and stress. The computational simulations of the material models are performed using the quasi-Monte Carlo (QMC) method with the design of random input variables by the Latin Hypercube Sampling (LHS). It is shown that the proposed approach can bring significant performance and dimensionality reduction to computational problems dealing with the modeling of heterogeneous materials. This strategy is well suited for the probabilistic approaches used in material modeling.

Czech name

—

Czech description

—

Type

O - Miscellaneous

CEP classification

—

OECD FORD branch

20101 - Civil engineering

Project

<a href="/en/project/TH75020002" target="_blank" >TH75020002: Deep-Learning-Enabled On-Demand Design of Composite Microstructure: Application to Mechanical Metamaterials</a><br>

Continuities

P - Projekt vyzkumu a vyvoje financovany z verejnych zdroju (s odkazem do CEP)

Publication year

2023

Confidentiality

S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů