Damage tolerance of impacted curved panels

Abstract

The final aim of this study is to evaluate the influence of impact damage on the residual strength of carbon/epoxy vessels stressed by internal pressure. An intermediate stage determined the residual behaviour of pre-impacted curved panels loaded in tension. Curved panels were impacted, reproducing the damage types observed in impacted vessels filled with propellant. Delamination damage was assessed by ultrasonics and optical microscopy used to observe intra-laminar mechanisms. Tension after impact (TAI) tests quantified the residual behaviour. An experimental design was used as an alternative to the complex analytical modelling of dynamic damage mechanisms. With this original technique, empirical relationships were established, linking impact parameters to residual properties. The force to failure was found to vary in a bi-linear manner with impact energy. Below a specific level of impact energy corresponding to failure in 4/7 of the plies, there is no significant reduction in the residual strength. The composite Young's modulus decreased linearly with impact energy.

Introduction

Composite structures can be accidentally damaged by dropping tools, or in collisions with foreign objects in manufacturing, storage or operation. Although it is easy to detect this kind of damage in metallic structures, it is not in composite structures, particularly when the reinforcement used is carbon fibre. Therefore, since it is difficult to avoid accidents, it is necessary to evaluate the effects of damage. There are a number of damage tolerance studies for this kind of material [4], [9], [15], [16], [2], [14]. The procedure used consists of three steps: (i) damage initiation, either using real conditions or reproducing them as closely as possible, (ii) damage inspection, evaluating its nature and topology, and (iii) the quantification of the residual strength, when the structure is stressed under simulated real conditions. Having determined the maximum structural damage that can be tolerated for a given residual strength, there is often a need for structural re-design.

The aim of our study is to evaluate the influence of damage on the residual strength of carbon/epoxy vessels. This generally requires a great number of test results. Study of the real structures is impossible because of the excessive cost involved. Hence specimens are extracted from these structures. In this case, curved panel specimens are extracted from tubes.

Although the literature contains papers concerning the residual burst strength of impacted composite tubes [6], [10], few studies have focused on damage tolerance of composite curved panels, especially when loaded in tension along the longitudinal axis. Our study is a first step to understanding the physical phenomena which take place when shear or normal stresses (tensile or compressive) are induced at the mesoscale of the multilayered composite. The main damage mechanisms observed, in impacted vessels filled with propellant, are fibre failure and localized delamination. Since fibre failure is the most critical damage mechanism for the tension strength of composite specimens, defect initiation tests were chosen to generate this damage mode and thus cause a significant loss of strength during quasi-static tension tests. Delamination is deliberately limited here to isolate the influence of fibre failure on the residual strength.

Two different methods can be used to predict structural damage due to impact. The first is semi-analytical modelling, in which the interactions between damage mechanisms (fibre failure, delamination …) and the coupling between local damage and the global response of the structure, are considered. However, the boundary conditions and the contact conditions in impact tests are difficult to evaluate. Since the phenomena occur on a millisecond timescale, it is only possible to identify them experimentally by doing interrupted tests. This strategy was not used because of the small number of samples available.

The second modelling method is empirical using an experimental design [11]. This “black-box” model links quantities, representative of damage, to impact parameters. Impact kinetic energy was not chosen as an input, since very different modes of failure can correspond to the same kinetic energy, according to the mass–velocity couple [11]. Therefore, we chose to consider the mass and velocity as independent parameters. It is possible to explore a parameter domain, consistent with the real conditions, while minimizing the number of tests required.