Uncorrelated volume element for stochastic modeling of microstructures based on local fiber volume fraction variation

Seyed Hamid Reza Sanei, Ray S. Fertig

Research output: Contribution to journalArticle

30 Citations (Scopus)

Abstract

One source of microstructural variability in fiber reinforced polymers is variability in fiber volume fraction. Despite large scatter in fiber volume fraction, an average value is usually reported without any consideration regarding the variability. Significant variation in fiber volume fraction across the sample reveals that homogeneity throughout the sample is a poor assumption and the commonly employed representative volume element is a poor representation of real microstructures. In this work, the distributions of fiber volume fraction at different length scales are investigated. The variation in distributions suggests that a length scale-dependent distribution must be selected in probabilistic analysis. A cross-correlation between fiber volume fractions of nearest neighbors was computed and an uncorrelated volume element (UVE) is introduced for the length scale at which fiber volume fractions become uncorrelated. The concept of a UVE enables random attribution of properties in stochastic modeling.

Original languageEnglish (US)
Pages (from-to)191-198
Number of pages8
JournalComposites Science and Technology
Volume117
DOIs
StatePublished - Sep 9 2015

Fingerprint

Volume fraction
Microstructure
Fibers
Polymers

All Science Journal Classification (ASJC) codes

  • Ceramics and Composites
  • Engineering(all)

Cite this

@article{557a01f51b194b77818125f4f399687e,
title = "Uncorrelated volume element for stochastic modeling of microstructures based on local fiber volume fraction variation",
abstract = "One source of microstructural variability in fiber reinforced polymers is variability in fiber volume fraction. Despite large scatter in fiber volume fraction, an average value is usually reported without any consideration regarding the variability. Significant variation in fiber volume fraction across the sample reveals that homogeneity throughout the sample is a poor assumption and the commonly employed representative volume element is a poor representation of real microstructures. In this work, the distributions of fiber volume fraction at different length scales are investigated. The variation in distributions suggests that a length scale-dependent distribution must be selected in probabilistic analysis. A cross-correlation between fiber volume fractions of nearest neighbors was computed and an uncorrelated volume element (UVE) is introduced for the length scale at which fiber volume fractions become uncorrelated. The concept of a UVE enables random attribution of properties in stochastic modeling.",
author = "Sanei, {Seyed Hamid Reza} and Fertig, {Ray S.}",
year = "2015",
month = "9",
day = "9",
doi = "10.1016/j.compscitech.2015.06.010",
language = "English (US)",
volume = "117",
pages = "191--198",
journal = "Composites Science and Technology",
issn = "0266-3538",
publisher = "Elsevier BV",

}

TY - JOUR

T1 - Uncorrelated volume element for stochastic modeling of microstructures based on local fiber volume fraction variation

AU - Sanei, Seyed Hamid Reza

AU - Fertig, Ray S.

PY - 2015/9/9

Y1 - 2015/9/9

N2 - One source of microstructural variability in fiber reinforced polymers is variability in fiber volume fraction. Despite large scatter in fiber volume fraction, an average value is usually reported without any consideration regarding the variability. Significant variation in fiber volume fraction across the sample reveals that homogeneity throughout the sample is a poor assumption and the commonly employed representative volume element is a poor representation of real microstructures. In this work, the distributions of fiber volume fraction at different length scales are investigated. The variation in distributions suggests that a length scale-dependent distribution must be selected in probabilistic analysis. A cross-correlation between fiber volume fractions of nearest neighbors was computed and an uncorrelated volume element (UVE) is introduced for the length scale at which fiber volume fractions become uncorrelated. The concept of a UVE enables random attribution of properties in stochastic modeling.

AB - One source of microstructural variability in fiber reinforced polymers is variability in fiber volume fraction. Despite large scatter in fiber volume fraction, an average value is usually reported without any consideration regarding the variability. Significant variation in fiber volume fraction across the sample reveals that homogeneity throughout the sample is a poor assumption and the commonly employed representative volume element is a poor representation of real microstructures. In this work, the distributions of fiber volume fraction at different length scales are investigated. The variation in distributions suggests that a length scale-dependent distribution must be selected in probabilistic analysis. A cross-correlation between fiber volume fractions of nearest neighbors was computed and an uncorrelated volume element (UVE) is introduced for the length scale at which fiber volume fractions become uncorrelated. The concept of a UVE enables random attribution of properties in stochastic modeling.

UR - http://www.scopus.com/inward/record.url?scp=84935112708&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84935112708&partnerID=8YFLogxK

U2 - 10.1016/j.compscitech.2015.06.010

DO - 10.1016/j.compscitech.2015.06.010

M3 - Article

AN - SCOPUS:84935112708

VL - 117

SP - 191

EP - 198

JO - Composites Science and Technology

JF - Composites Science and Technology

SN - 0266-3538

ER -