The Relationship between Linear Loop Density and Saturation Crack Density

In the crack density/strain plots shown in Figures 4.14 to 4.15, the upper and lower bounds for the linear loop density are also shown.

For the NCFA unpolished specimens (Figure 4.14), the NCFA upper bound linear loop density corresponds very closely to the experimentally observed saturation crack density.

In addition, the final crack density prior to laminate failure for the NCFA polished specimens shown in Figure 4.14.b, lies between the two bounds (note that saturation of damage does not occur in the polished specimens). A close inspection of edge sections of these specimens showed that in the unpolished specimens, the blankets were largely out- of-phase and hence the saturation crack density tended to the upper bound value. By contrast, the NCFA specimens shown in Figure 4.14b had blankets which were more in- phase.

In the NCFB specimens there is a different relationship between the crack density and linear loop density. Here, all specimens tended to behave in a similar manner (Figure 4.15), and the saturation crack density is much greater than the upper bound linear loop density of 0.6 mm'1. This is consistent with the lack of influence of the loops in the crack accumulation for these laminates, which have transverse fibres that are more evenly dispersed. However, some small influence of the loops on crack accumulation may be apparent from the figures since the crack density/applied strain results appear to show a change of slope at the upper bound line.

4.9. SUMMARY

Damage development in non-crimp fabric glass/epoxy cross-ply laminates has been studied under quasi-static tensile loading conditions. Two laminate lay-ups were studied, NCFA (WARP/WEFT), and NCFB (WEFT/WARP),. A characterisation of the transverse ply matrix cracking was carried out. Damage morphology and laminate microstructure was observed in detail and the development of damage quantified as a function of strain using both photography and microscopy. The relationship between laminate linear loop density and saturation crack density was also compared.

Stress-strain plots show a characteristic knee in the curve, corresponding to the onset of cracking, below which the stress/strain relationship is linear. Damage initiates at a lower strain for NCFA coupons than in NCFB coupons. Unpolished-edge specimens show damage initiation at a lower strain than polished-edge specimens, which is attributed to the increase in damage sites along the rough unpolished specimen edge.

The NCFA coupons show a slight variability in the transverse ply thickness, since nesting of the well-defined 90° ply fibre bundles is possible. In general, NCFB specimens have a thicker 90° ply due to the wavy, dispersed nature of the bundles in the transverse ply with no nesting observed. Although thick transverse plies (as in NCFB) would normally yield lower strain to crack initiation, this was not found to be the case and was attributed to the dispersed nature of the 90° fibre bundles and consequent lower strain magnification between tows.

The Young’s modulus was similar for all specimen types, with the slightly lower Young’s modulus recorded for the NCFA laminates being attributed to the waviness of the 0° tows.

Comparisons with predictions of the Young’s modulus based on rule-of-mixtures and laminate plate theory were in reasonable agreement with experimental results, although neither method took into account structural features within each laminate, since the predictions are based on an idealised, homogeneous cross-ply laminate. Failure strains and tensile strengths are much lower in the NCFA specimens, a result of the misalignment

of the 0° fibres by up to 10°.

Damage accumulation was observed to behave in a similar manner to that observed in cross-ply laminates. Cracks generally span the full width, with damage initiating at the edge in all cases as is usually the case for cross-ply GFRP laminates with ‘thick’

transverse plies. NCFA specimens with unpolished edges show damage propagating primarily through the loops, whereas this is not necessarily the case for polished edges.

In the NCFB laminates, there is no preference for propagation through the loops, and some cracks are halted by overlapping with cracks propagating from the opposite edge.

Thus the dispersed nature of the fibres in the transverse tows of the NCFB specimens dominates the development of damage. Cracks were observed in the surface resin-rich region for all specimen types, probably due to a combination of low fibre volume and fibre bundling, whereby it is sometimes possible for transverse ply cracks to find a path to the surface.

Saturation of matrix cracking damage occurred in the unpolished NCFA specimen, indicating that all the preferred sites associated with the loops were used. However, no saturation occurred in the polished type, as cracks were not confined to the loops. The NCFB specimens yielded higher strains to failure as a result of the good alignment of the 0° plies. No saturation of matrix cracking occurred in these specimens since, again, the matrix cracking was not confined to paths through or adjacent to loops.

For both the NCFA and NCFB laminate types, measurements of Young’s modulus during incremental loading showed an initial increase in modulus prior to a decrease once cracks began to form. The origin of the initial modulus increase is not entirely clear but may be related to straightening of the 0° tows. The degradation in modulus as a result of the accumulation of matrix cracks occurs in a similar manner to that observed in cross-ply laminates.

A new parameter, the linear loop density, has been introduced. This is defined as the number of loops per unit length of fabric blanket. In some circumstances, the upper bound linear loop density for a laminate correlated well with the saturation matrix crack

density. This occurs when the two fabric blankets are out-of-phase and the matrix crack initiation sites are associated with the resin-rich regions caused by the loops, as in the case of the unpolished NCFA coupons. The NCFB specimens all showed crack densities at failure well above the upper bound linear loop density, consistent with the lack of influence of the loops on crack accumulation, since the transverse tows in these coupons are more dispersed.

CHAPTER 4