Schmlicker et al / Composites: Part A 34(2003)613-622 a b Fig. 8. Three-dimensional distribution of the flaws shown in Fig. 7. Type i(a)and type iii flaws (c)change their x-position according to the fiber rovings winding angle, while type ii flaws(b)run throughout the sample without variations in the x-z-plane. For clarity reasons only one kind of each flaw type is shown Type ii flaws. This type of flaws does not follow the the slurry infiltrated fiber bundles. Drier rovings, however, winding angle as one moves in y-direction from one slice to are less flexible and formable and that way they are less apt the other. Consequently, a direct correlation with the fiber at settling into place, adapting the right shape and alignment cannot exist. This type of flaws may be correlated combining to laminae with the winding procedure then used, delivering period- Type iii faws. These are matrix-rich areas, which are ically slightly different states of drying prior to winding of however, not completely fiber free. Probably these flaws correspond to one single roving, that has been spread out In the meantime the winding technique has been improved in a way that rustically. This interpretation is supported by the corre drying is carried out homogeneously through the whole winding process. lation of the three-dimensional distribution of this flawType ii flaws. This type of flaws does not follow the winding angle as one moves in y-direction from one slice to the other. Consequently, a direct correlation with the fiber alignment cannot exist. This type of flaws may be correlated with the winding procedure then used1 , delivering period￾ically slightly different states of drying prior to winding of the slurry infiltrated fiber bundles. Drier rovings, however, are less flexible and formable and that way they are less apt at settling into place, adapting the right shape and combining to laminae. Type iii flaws. These are matrix-rich areas, which are, however, not completely fiber free. Probably these flaws correspond to one single roving, that has been spread out drastically. This interpretation is supported by the corre￾lation of the three-dimensional distribution of this flaw Fig. 8. Three-dimensional distribution of the flaws shown in Fig. 7. Type i (a) and type iii flaws (c) change their x-position according to the fiber rovings’ winding angle, while type ii flaws (b) run throughout the sample without variations in the x–z-plane. For clarity reasons only one kind of each flaw type is shown. 1 In the meantime the winding technique has been improved in a way that drying is carried out homogeneously through the whole winding process. 618 M. Schmu¨cker et al. / Composites: Part A 34 (2003) 613–622