Discussion of the statistical representativeness of the results in: Lomov, Breite, Melnikov, Mesquita, Swolfs and Abaimov [Int. J. Solids Struct 225 (2021) 111061]

Modelling on stress redistribution around broken fibers for fiber reinforced plastic with poor fiber distribution uniformity

“…Based on FE modelling, Blassiau et al [14,26] established the representative volumes were 16, 32, and 36 intact fibers are regularly arranged near broken fibers, and the nearby stress distribution together with fiber damage is calculated. Lomov et al [27] combined FE modelling and numerical methods to calculate SCF and progressive fiber breaking number, for FRP with random fiber distribution. Swolfs et al [22,25] used FE modelling on representative volumes containing 40, 80, and 120 fibers to calculate the SCF, where the random fiber distribution is considered, and an analytical SCF model is proposed based on the FVR and the distance between intact and broken fibers.…”

“…[22,24,[29][30][31][32] Although such modelling provides good and realistic deformation compatibility between fiber and matrix, the small diameter of fibers and very small distance between adjacent fibers require fiber and matrix elements in a very small size, which leads to much computation time if many fibers are included in FE models. Meanwhile, although current studies have taken into account practical random fiber distribution, the fiber distribution is quite uniform, [22,25,27,33,34] and SCF in regions of quite non-uniform fiber distribution such as resin-rich pockets or fiber clusters [35][36][37] is barely addressed. If random fiber distribution with poor uniformity is to be investigated by FE modelling, much more fibers needs to build and the computation efficiency of FE calculation needs to be improved.…”