Blast Testing and Modelling of Composite Structures

The motivation for this work is based on a desire for finding light weight alternatives to high strength steel as the material to use for armouring in military vehicles. With the use of high strength steel, an increase in the level of armouring has a significant impact on the vehicle weight, affecting for example the manoeuvrability and top speed negatively, which ultimately affects the safety of the personal in the vehicle.

Strong and light materials, such as fibre reinforced composites, could therefore act as substitutes for the high strength steel, and minimize the impact on the vehicle weight or actually reduce the weight. Before such materials can be brought into use, their performance against blast loading need to be evaluated.

This can be done through the use of small scale blast testing. The overall objectives of this thesis have therefore been to establish and validate an experimental facility, usable for performing small scale blast test on laminate and sandwich panels, and to set-up a numerical framework for modelling the test panel response when impacted by a blast load.

The test set-up was designed such that the panel response could be measured by use of high-speed DIC (Digital Image Correlation). A test series, using Eglas/Epoxy laminates and sandwich panels, with Eglas/Epoxy skins and foam core, was conducted, to evaluate the functionality of the designed test set-up, and to gain insight into the response of the panels, when impacted by a blast load.

The test set-up proved functional and provided consistent data of the panel response. The tests reviled that the sandwich panels did not provide a decrease in panel deflection compared with the monolithic laminates, which was expected due to their higher flexural rigidity. This was found to be because membrane effects became the controlling parameter for the panel deflection, activated by the large deflection of the panels relative to their thicknesses.

The tests on the sandwich panels showed that no compression of the core had taken place, an effect that was thought could be utilized for absorbing energy from the blast pressure, but which had to be rejected. A comparison between E-glas/Epoxy and S-glas/Phenol laminates, with a quarto-axial (QA) and plain weave (PW) fibre layup respectively, showed that the S-glas/Phenol system could, as a minimum, withstand the load from an explosive charge 50% larger than the E-glas/Epoxy laminate could, without rupturing, indicating that the PW layup has attractive properties for absorbing energy from a blast load.

To model the blast tests the numerical solver ls-dyna was used. The blast load was modelled using two approaches; (i) with *load_blast_enhanced model in ls-dyna, which applies a pressure distribution on a selected surfaces and has been based on experimental pressure measurement data, and (ii) with a designed 3 step numerical load model, where the blast pressure and FSI (Fluid Structure Interaction) between the pressure wave and modelled panel is modelled numerically.

The tested laminates and sandwich panels was modelled using material models available in ls-dyna. Comparison between modelled and tested panel response from a 25g charge detonated 100mm from the panel surface, showed the modelled panel response to be 19% lower than thetest data. This difference could be argued to originate from test set-up uncertainties, but also due to inconsistencies between model and test pressure.

It was attempted to design a test set-up to measure the blast pressure, but the variation in the measured pressure data was too large to be used for comparison with the modelled pressure. In a future work this set-up should be improved such that the modelled pressure can be validated. For tests performed with a 250g charge load comparisons with model data showed poor agreement.

This was found to be due to improper design of the modelled laminate panels, where the layer interface delamination was not represented physically, but was taken into consideration through failure parameters in the used laminate model. To improve the observed behaviour the layer interface should be modelled, by using a cohesive zone approach, based on fracture resistance data for the layer interfaces.

By comparing model and test data for the blast testing performed on the sandwich panels, it was found that in the models the foam core was compressed, a behaviour opposite what was identified from the test results. The models showed that the foam was compressed with a strain rate, several orders of magnitude larger than the strain rate used in the material testing performed to obtain data for describing the strain rate sensitivity of the foam.

Extrapolating the rate behaviour to such large strain rate might therefore not be valid, and the material testing should be extended to include tests at higher strain rates. Other reasons for the inconsistency are believed to be due to improper representation of the laminate interfaces and the skin core interface.