digilib@itb.ac.id +62 812 2508 8800

Abstrak - Gading Wulung Wiradjanu
Terbatas  Irwan Sofiyan
» Gedung UPT Perpustakaan

BAB 1 Gading Wulung Wiradjanu
Terbatas  Irwan Sofiyan
» Gedung UPT Perpustakaan

BAB 2 Gading Wulung Wiradjanu
Terbatas  Irwan Sofiyan
» Gedung UPT Perpustakaan

BAB 3 Gading Wulung Wiradjanu
Terbatas  Irwan Sofiyan
» Gedung UPT Perpustakaan

BAB 4 Gading Wulung Wiradjanu
Terbatas  Irwan Sofiyan
» Gedung UPT Perpustakaan

BAB 5 Gading Wulung Wiradjanu
Terbatas  Irwan Sofiyan
» Gedung UPT Perpustakaan

COVER Gading Wulung Wiradjanu
Terbatas  Irwan Sofiyan
» Gedung UPT Perpustakaan

DAFTAR PUSTAKA Gading Wulung Wiradjanu
Terbatas  Irwan Sofiyan
» Gedung UPT Perpustakaan

LAMPIRAN Gading Wulung Wiradjanu
Terbatas  Irwan Sofiyan
» Gedung UPT Perpustakaan

This undergraduate thesis investigates the modelling of transverse matrix crack initiation and evolution in thermoplastic fibre-reinforced polymer (FRP) composites using a multiscale approach known as the Direct FE² method. Direct FE² enables concurrent simulation of macro- and micro-scale behaviour by embedding Representative Volume Elements (RVEs) directly into the finite element formulation. A modified version of Direct FE2 is used where it can handle the case of multiple RVE in a randomized configuration. To capture the variability of composite microstructure, python script is developed to generate four unique RVEs with randomly distributed, non-overlapping fibres at a target volume fraction of 47%. These RVE are then placed randomly throughout the macroscale model using the modified Direct FE2. Two different macroscale models are studied: a single-element model and a 40-element specimen model based on ASTM standards. Cohesive zone modelling is implemented at the fibre-matrix interface to simulate debonding and matrix cracking. Numerical results from Direct FE² simulations are compared against existing experimental and numerical benchmarks. The simulations demonstrate good agreement in elastic and plastic behaviour, with the random RVE model yielding more accurate stiffness predictions than manually constructed RVEs.