Fast Crack Growth In Aircraft And Aeroengine Structures

Summary of Final Report

Prof. C. Ruiz, Dr D. Nowell and Dr N. Petrinic

The safety of a structure is normally assessed by postulating the existence of a flaw and studying its response to a static or slowly varying load.There are however many situations where dynamic effects must be considered such as impact loading followed by rapid crack propagation.These may arise when a flaw is suddenly introduced in a prestressed panel.Examples found in the aerospace industry range from the explosion of pressurised fuselages triggered by a fatigue flaw or a hole punched by a missile to the non-containment of a burst aeroengine disk or a blade.

An experimental investigation into the crack propagation in aluminium alloy 2014 has been completed.Panels 120 mm wide by 300 mm long were subjected to a constant force applied by two identical pneumatic cylinders and a crack was induced by forcing open a pre-machined notch with a wedge travelling at approximately 25 m/s.The crack growth was observed by high speed photography and the state of stress of the test piece was monitored by means of strain gauges.Crack arresters in the form of slots and ribs have been studied.A dynamic photoelastic experiment was also used to obtain the elastic stress intensity factor.The results have highlighted the effect of the rolling direction of the material.

Two finite element codes have been used to analyse and interpret the results: ABAQUS and DYNA.The following conclusions were drawn from the experiments.

1. The process of crack growth in the plastic domain is fundamentally different from the elastic domain. L.E.F.M. criteria such as the stress intensity factor are not applicable to the ductile failure found in the aluminium panels tested.
2. The potential of simple criteria such as the Crack Opening Angle in combination with the F.E. code ABAQUS has been explored.It is concluded that this approach can be used to extrapolate from a small scale experiment to a real situation but that the experiment needs to be carefully designed in order to reproduce the main features of the real situation.
3. Damage mechanics has proved particularly effective in interpreting the experimental observations.The use of a simplified material failure model, based on the Sandia model implemented in the F.E. code DYNA 3D, although successful when dealing with the panels tested in this investigation, presents practical difficulties for general application.These are related with the number of elements required, the computational power needed in view of the iteration times and the relative complexity of even the simplest failure model that can be devised compatible with an acceptable accuracy.
4. While most of these practical difficulties will be overcome as greater computing power becomes available, there still remains the fundamental problem associated with the uncertainty in the values of the material parameters that must be defined within the chosen material failure model.In this investigation these parameters have been deduced from the tests.Further work is needed to measure them directly from simple material tests.
5. The next step is to devise tests which, in combination with DYNA-3D analysis, will provide a direct measurement of the material parameters required by the damage mechanics failure model.In addition to ultra-high speed photography in combination with moire interferometry, reflection photoelasticity and strain gauges, the temperature will be monitored using the fast response infrared apparatus developed in Oxford.

This project was funded by the EPSRC (Grant Reference GR/L32453)