Discussions on extension of traditional cohesive element for delamination modeling of laminates used in combination with phantom node intraply elements

Abstract

Discrete damage modeling techniques based on phantom node elements are promising for failure study of composite laminates. The necessity of extending cohesive element in order to match phantom node solid elements has been discussed by several researchers but no consensus has been arrived. Moreover, although several extended cohesive elements have been proposed, damage variable transfer between new and original integration points upon subdomain partition has seldom been discussed. In the current study, the theoretical significance of extending cohesive element is demonstrated focusing on modeling of delamination initiation induced by matrix cracks. Then, a phantom node cohesive element is introduced and an approximate damage variable transfer algorithm is proposed. Validity of the extended interface element as well as the damage variable transfer algorithm is investigated under practical situations and comparisons are made between modeling results in order to further investigate the necessity of extending cohesive element. According to the current research, the necessity of extending cohesive element and of developing damage variable transfer algorithm are not as important as expected for quasi-static failure modeling of notched laminates. However, in order to deal with more complex situations, further investigations are still in need.