Delamination of Composite Blade Structures using the Cohesive Zone Approach

Currently, the design of wind turbine blades heavily relies on extensive testing and a very conservative design philosophy namely "Damage resistant" designs when it comes to the assessment of delamination. Damage resistant designs refer to designing a blade in a way that it resists damage formation, which means that no new damage will occur, and the existing damage will not propagate during the service life of a wind turbine. This, however, does not match observations as wind turbines do show damage formation (both initiation and propagation), even to the point of collapse. Overly conservative design philosophies can result in thicker and heavier structures than needed. This is not only inefficient but it can defeat its very same objective considering that one of the main loads of large wind turbine blades is the weight of the blade itself. There is an alternative less conservative approach namely "Damage tolerant" design philosophy, which, contrary to damage resistant designs, allows the initiation and the slow propagation of damage until it reaches some critical propagation. It is uncontroversial to say that damage tolerant design would produce better, more efficient wind turbines by dramatically increasing the design space. However, to implement damage tolerant designs there need to be tools that accurately predict the propagation of delamination of structures under realistic loading conditions. The present thesis aims at developing such tools in the form of a cohesive zone model that is capable of describing delamination of composites under mixed-mode loading and accounts for the main toughening mechanism of unidirectional composites i.e. fibre bridging.

The development of the tool consisted in formulating a new mixed-mode cohesive law. This cohesive law was formulated using a coupled potential function, which was determined experimentally using R-curves. In this way, it can be said that the cohesive law is "measured" for a given material interface. The derived cohesive law was used for the characterisation of both the crack tip and the bridging fracture of a unidirectional glass/epoxy composite. The derivation of the cohesive law and the experimental characterisation of the UD glass/epoxy composite are presented in detail in Paper A, and Paper B (attached in Chapter 9, and 10), respectively. In the papers, a methodology is proposed and explained in order to derive the specific cohesive law of other material based on R-curves (expressed in terms of the opening, and the J-integral). As part of the Ph.D. work, this cohesive law has been implemented into a cohesive zone model (CZM), which can be used in commercial finite element software (ABAQUS for the present work). The CZM was implemented via a subroutine that defines the traction separation response and the constitutive stiffness matrix of a predefined cohesive element (COH2D4, and/or COH2D6). The subroutine was subject to different 1-element tests to verify its ability to re-create the defined cohesive law at different stages and conditions. The functionality and robustness of the subroutine were corroborated. The implemented subroutine showed that it is capable to model the crack tip tractions as well as the bridging traction under mixed-mode loading, unloading, and reloading states.