Co-authors​ :

 Bent F. SØRENSEN (DENMARK), Frode GRYTEN (NORWAY), Reidar K. JOKI (NORWAY), Jens K. JØRGENSEN , Philippe NOURY (NORWAY), Daan J. H CEDERLØF (DENMARK) 

Failure of composite structures often involves delamination, crack growth between layers. For some composite materials, denoted damage tolerant materials, the fracture resistance increases with crack length (R-curve behaviour). The underlying physical mechanism is large-scale bridging by fibres that cross over the crack and create tractions along the crack faces opposing the crack opening. The fracture resistance, characterized in terms of the J integral [1], rises from an initial value, J_0, (a sharp, unbridged crack) to a significantly higher steady-state value, J_ss, (fully developed bridging zone) under monotonic loading [1]. Under cyclic loading, the crack growth rate is fastest in the beginning (sharp, unbridged crack) and decreases as the bridging zone becomes larger, and reaches a significantly slower growth rate as the bridging zone is fully evolved [2]. From an engineering perspective, such material behaviours are very attractive. However, current design approaches based on linear elastic fracture mechanics do not utilize these behaviours. This paper proposes a new design approach to utilize the advantageous properties of damage tolerant composites, shifting from a 'no crack growth' to a 'controlled crack growth' philosophy, allowing for stable but not unstable crack growth.

We will consider three types of design approaches: Level 1, Level 2 and Level 3. Level 1 is a traditional design where the applied J integral value, J should be kept below J_0, so that no crack growth occurs, and cyclic crack growth is assessed by the Paris-Erdogan relation for unbridged crack. Level 2 is restricted to a class of problems we call steady-state problems for which the J integral is independent of crack length, so that under monotonic loading unstable cracking will not occur unless J reaches J_ss. The analysis can be made using analytical models. Level 3 is design using cohesive laws. Level 3 is not restricted to steady-state problem. However, Level 3 is also more demanding; the cohesive laws of the failure plane must be derived from the measured fracture resistance and a numerical model (e.g. a finite element model) must be used in the analysis. Thus, we assess that the Level 2 design approach would be easier to bring into engineering design. The paper will illustrate the Level 2 approach by analysing delamination crack growth from a ply-drop under static and cyclic loading, identified as a steady-state problem. We will outline the steps in this design approach, demonstrating that the Level 2 design stress for both static and cyclic crack growth can be significantly higher than that for Level 1. Finally, we will discuss the implications regarding safety factors and the limitations of our proposed approach.