M. Schmiicker et al /Composites: Part A 34 (2003)613-622 3.3. Correlation between mesostructural flaws and shear The ILSs of 20 samples taken from the investigated WHiPOX plate that was mildly compressed in the moist stage( 10% thickness reduction) scatters between s 4 and =13 MPa(Fig. 10a). Obviously, there is no correlation between ILSs values and the location of samples in the plate. The comparison of shear strengths of samples I and l (12 and =7 MPa, respectively) and the mesostructural analyzes suggest that in a first order approximation a correlation exists between flaw population and shear strength. The percentage flaw area of sample I(12 MPa) of sampleⅡ(≈7MPa)is9.9% 2 However. as we will see below, the shear strength is controlled by local flaw concentrations (i.e. pronounced interlaminate matrix enrichments) rather than by the Winding process causes percentage of fiber free areas alone The summed-up projections of the flaw population onto local misalignments of the z-axis(Fig 10c, see also Fig. 5) clearly show that the infiltrated fiber rovings flaws are concentrated in planes parallel to the fiber laminae It can be assumed that the interlaminate matrix agglomera- Plane 2 tions, which typically contain large pores, act as weak points Plane 1 planes of delamina The flaw population does not only vary from sample to sample but also from slice to slice within one sample (Fig. 10b). Especially in sample l, both, slices with large and By consolidation in the wet small fiber free areas occur. as the total number of fibers stage parallel rovings form the sample is taken to be constant, the fibers missing in flaw laminae areas either are concentrated in other parts of the sample, or the sample sizes are enlarged locally to accommodate the holes. The fiber cell analysis(see Fig. 6) of micrograph from flaw-rich and flaw-poor slices yields an average cell size of the latter which is somewhat greater(200 um) than that of the flaw-rich slices(170 um). This indicates that the slices with large fiber free areas must have Misaligned rovings form increased fiber density within the fiber-rich regions. In order to understand the influence of the processing distorted laminae or wedge parameters on the distribution of flaws and related ILss some shaped breakoff, both fabrication-related details have to be taken into account: after leading to fiber-free zones the winding process the material is removed from the mandrel, flattened, and then pressed between two Al2O plates in order to achieve greater density and better homogeneity. To determine the influence moist-stage com- pression has on the mesostructure, a strongly compressed Fig9.Schematic illustration showing the origin of fiber free areas caused sample(sample ml, about 25% volume change)was by an angle a between winding direction and the fibers'direction. compared with mildly compressed materials(samples I and Il, about 10% volume change, see above). Fig. ll gives the type and the winding angle. As the rovings are wound at a ILSS data throughout the strongly compressed WHIPOX 15 angle, this type of flaw will naturally follow the same plate, together with the three-dimensional illustration of the angle. A possible explanation for the origin of such a flaw distribution of a selected sample, and the cumulative flaw roving defect is partial retention of fiber sizings during the frequencies projected on the z-axis ILSS data are somewhat wet stage. By that, complete infiltration of the fiber bundle higher and the distribution throughout the plate is more with matrix slurry is impeded and later-on voids are uniform in comparison to the mildly compressed WHIPOX formed during the burn-off of the relic sizing polymer plate(see Fig. 10). Accordingly, the flaw population is more during the sintering stage homogenous as evidenced by the uniform maxima oftype and the winding angle. As the rovings are wound at a 158 angle, this type of flaw will naturally follow the same angle. A possible explanation for the origin of such a roving defect is partial retention of fiber sizings during the wet stage. By that, complete infiltration of the fiber bundle with matrix slurry is impeded and later-on voids are formed during the burn-off of the relic sizing polymer during the sintering stage. 3.3. Correlation between mesostructural flaws and shear strength The ILSS of 20 samples taken from the investigated WHIPOX plate that was mildly compressed in the moist stage (<10% thickness reduction) scatters between <4 and <13 MPa (Fig. 10a). Obviously, there is no correlation between ILSS values and the location of samples in the plate. The comparison of shear strengths of samples I and II (<12 and <7 MPa, respectively) and the mesostructural analyzes suggest that in a first order approximation a correlation exists between flaw population and shear strength. The percentage flaw area of sample I (<12 MPa) is 4.5% while that of sample II (<7 MPa) is 9.9%. However, as we will see below, the shear strength is controlled by local flaw concentrations (i.e. pronounced interlaminate matrix enrichments) rather than by the percentage of fiber free areas alone. The summed-up projections of the flaw population onto the z-axis (Fig 10c, see also Fig. 5) clearly show that the flaws are concentrated in planes parallel to the fiber laminae. It can be assumed that the interlaminate matrix agglomera￾tions, which typically contain large pores, act as weak points and as probable planes of delamination. The flaw population does not only vary from sample to sample but also from slice to slice within one sample (Fig. 10b). Especially in sample I, both, slices with large and small fiber free areas occur. As the total number of fibers in the sample is taken to be constant, the fibers missing in flaw areas either are concentrated in other parts of the sample, or the sample sizes are enlarged locally to accommodate the ‘holes’. The fiber cell analysis (see Fig. 6) of micrographs from flaw-rich and flaw-poor slices yields an average cell size of the latter which is somewhat greater (<200 mm2 ) than that of the flaw-rich slices (<170 mm2 ). This indicates that the slices with large fiber free areas must have an increased fiber density within the fiber-rich regions. In order to understand the influence of the processing parameters on the distribution of flaws and related ILSS some fabrication-related details have to be taken into account: after the winding process the material is removed from the mandrel, flattened, and then pressed between two Al2O3 plates in order to achieve greater density and better homogeneity. To determine the influence moist-stage com￾pression has on the mesostructure, a strongly compressed sample (sample III, about 25% volume change) was compared with mildly compressed materials (samples I and II, about 10% volume change, see above). Fig. 11 gives the ILSS data throughout the strongly compressed WHIPOX plate, together with the three-dimensional illustration of the flaw distribution of a selected sample, and the cumulative flaw frequencies projected on the z-axis. ILSS data are somewhat higher and the distribution throughout the plate is more uniform in comparison to the mildly compressed WHIPOX plate (see Fig. 10). Accordingly, the flaw population is more homogenous as evidenced by the uniform maxima of Fig. 9. Schematic illustration showing the origin of fiber free areas caused by an angle a between winding direction and the fibers’ direction. M. Schmu¨cker et al. / Composites: Part A 34 (2003) 613–622 619