Lee et al. (2009) — Quantitative PVA Fiber Dispersion Evaluation

Citation

Bang Yeon Lee, Jin-Keun Kim, Jeong-Su Kim, Yun Yong Kim (2009). Quantitative evaluation technique of Polyvinyl Alcohol (PVA) fiber dispersion in engineered cementitious composites. Cement & Concrete Composites, 31(6), 408–417.

DOI: 10.1016/j.cemconcomp.2009.04.002
Atlas layer: supporting
Related Victor Li book chapter: Chapter 5: Processing and Rheology of ECC (also Chapter 4)
Source PDF: primary_data/lee-2009-quantitative-evaluation-technique-of-polyvinyl.pdfIJP00209E_Quantitative evaluation technique_1_CCC.pdf`
Extracted text: secondary_data/full_texts/lee-2009-quantitative-evaluation-technique-of-polyvinyl_full_text.mdsecondary_data/full_texts/IJP00209E_Quantitative evaluation technique_1_CCC_full_text.md`
Source note: secondary_data/source_notes/lee-2009-quantitative-evaluation-technique-of-polyvinyl_source_note.mdsecondary_data/source_notes/IJP00209E_Quantitative evaluation technique_1_CCC_source_note.md`

Why this paper matters

Solves the critical problem of evaluating low-contrast organic PVA fiber dispersion in cementitious composites. It establishes a quantitative methodology using fluorescence microscopy and an ANN-watershed-morphological image processing algorithm to calculate the fiber dispersion coefficient ($\alpha_f$).

Main contribution

GFP Fluorescence Detection: Utilized Green Fluorescent Protein (GFP) filtering under mercury lamp illumination to make non-conductive, low-contrast PVA fibers clearly visible as green cross-sectional dots.
Enhanced Segmentation Algorithm: Overcame the over-segmentation of single elongated fibers and under-segmentation of fiber bundles using a 5-feature ANN classifier (95.3 % accuracy) combined with morphological reconstruction.
Unit-Size Independent $\alpha_f$ Evaluation: Formulated a robust calculation for the fiber dispersion coefficient $\alpha_f$, validating it on benchmark images ($\alpha_f = 1.0$ for perfect, $0.041$ for severely biased, and $0.348$ for real ECC).

Evidence summary

Classifier Accuracy: 95.3 % on 1350 fiber database using a 5-4-4 Levenberg-Marquardt ANN (Table 3, Page 412).
Fiber Minor Axis Measurement: Average minor axis 42.2 $\mu\text{m}$ (6.84 % error from nominal 39.5 $\mu\text{m}$) (Table 4, Page 413).
Real Cross-Section Dispersion: Real composite cutting plane evaluated at $\alpha_f = 0.348$ across 565 detected fibers (Table 5, Page 415).

Linked Atlas nodes

02_concepts/fiber_dispersion.md
02_concepts/processing_rheology.md
02_concepts/fiber_bridging_law.md

Relationship to Victor Li book

Primary book anchor remains Victor Li (2019), Engineered Cementitious Composites (ECC).
This paper directly supports Chapter 5 (Processing & Rheology) and Chapter 4 (Material Design) by providing the empirical tool to measure the fiber dispersion coefficient ($\alpha_f$) that governs the fiber bridging performance ($\sigma_0$).

Claim-evidence rows to add

Atlas node	Claim	Evidence summary	Page/Figure/Table	Status
`02_concepts/fiber_dispersion.md`	Automated GFP fluorescence image analysis accurately determines the PVA fiber dispersion coefficient $\alpha_f$	Calculated $\alpha_f = 1.0$ for uniform, $0.041$ for clustered, and $0.348$ for real ECC cross-section	Page 408 / Eq. (1) / Table 5	verified_from_pdf
`02_concepts/fiber_dispersion.md`	A 5-feature ANN combined with morphological reconstruction eliminates over-segmentation with 95.3 % classification accuracy	5-4-4 ANN categorized 1350 fiber shapes into single and aggregated clusters with 95.3 % accuracy	Page 412 / Table 3 / Fig. 13	verified_from_pdf
`02_concepts/fiber_bridging_law.md`	Measured PVA fiber dispersion in real ECC cutting planes exhibits non-ideal distribution ($\alpha_f \approx 0.35$)	Whole image analysis of 565 fibers yielded an $\alpha_f$ of 0.348	Page 415 / Table 5	verified_from_pdf

Verification status

PDF preserved: yes (in primary_data/IJP00209E_Quantitative evaluation technique_1_CCC.pdf)
Text extracted: yes (PyMuPDF, 10 pages)
DOI verified: yes (10.1016/j.cemconcomp.2009.04.002)
Page/figure/table verified: yes (all checked in PDF text)
Claim-evidence matrix ready: yes

Cautions

Do not use this paper as evidence for composite tensile strain capacity or compressive strength, as mechanical stress-strain testing was not conducted in this specific publication.
Sizing of fibers carries a ~6.8 % optical enlargement error due to threshold selection.