Abstract. Based on a previous optimisation of set-up parameters for injection moulding of polyamide 6-6 (PA 6-6) reinforced with 40 wt% of 10 mm long glass fibre, the aim of this paper is to define suitable guidelines to further improve the mechanical performances of PA 6-6/glass long fibre thermoplastic (LFT) injection mouldings. Different solutions have been considered: screw and non-return valve design modification so as to adapt them to LFT processing, increase of the initial fibre content (up to 50 wt%) and length (up to 25 mm) in the LFT pellets. Using a LFT dedicated plasticating unit has allowed decreasing the fibre breakage amount by about 80% at the nozzle exit, however without improving the flexural properties. Increasing the initial fibre content has logically permitted to improve the flexural properties. Increasing the initial fibre length has not brought any improvement of the properties and has even amplified the structural heterogeneities and anisotropy of the parts. These trends have been explained on the basis of cavity pressure records highlighting significantly different rheological behaviours, and of resulting residual fibre lengths and through-the-thickness fibre orientation profiles modifications.

This paper focuses on the development and the validation of flexural modulus and flexural strength predictive models of long glass fibre reinforced polyamide 6.6 (PA66). Based on previous injection moulding optimization of 40 wt% long glass fibre PA66, a microstructure analysis was investigated on glass fibre reinforced PA66 by varying the parameters of the material (fibre length, fibre content, fibre diameter). In a first phase, analytical models established within the framework of the processing condition limits previously determined have been elaborated. These models lead to a good experimental/calculation correlation but remain limited to a mould and part design. In a second phase the flexural modulus and maximal flexural stress have been then estimated from structural models based on a five layer morphological description of the composites (local residual fibre length, local fibre content and fibre orientations). The long glass fibre PA66 composites were characterized in terms of fibre content distribution model and fibre orientation model through the part thickness. The experimental/model correlation was achieved whatever the process variability is (mould, material and processing conditions) both for the flexural modulus or flexural strength. The models have been then validated with an industrial part. Finally, a correlation between the two studied properties has been revealed depending on the nature of the composite matrix (PA66, PA6 or PP).

This article aims to investigate the flexural creep behaviour as a function of temperature of long glass fibre polyamide 6.6 taking into account the thermal-oxidative degradation occurring during the test. The mould geometry has been chosen so as to reproduce some geometrical accidents (e.g. sharp frontal and tangential steps) occurring on industrial moulds. The nominal fibre content (10, 40 and 55 wt%), initial fibre length (short glass fibre, long glass fibre), load rate (up to 70%) and creep temperature (23℃, 100℃ and 130℃) have been considered to estimate the Findley’s model coefficients. A first investigation on the polyamide 6.6 degradation under thermo-oxidative environment has been led to understand the mechanisms of thermal-degradation of the polyamide 6.6 composites. The pure polyamide 6.6 matrix has shown a 20% increase of flexural modulus during the first period of ageing attributed to a combined chain scissions and cross-linking reactions. Then, a decrease of properties attributed to predominant chain scission mechanism was noticed after 1000 h of thermal exposure reaching up to 30% after 5000 h. In case of reinforced polyamide 6.6, the flexural properties tend to increase (+6.5%) up to 2000 h of exposure. The least square method has then permitted to evaluate the material coefficients from the experimental data; the instantaneous creep strain has been estimated from a power law representation. In any cases, the calculations are in a good accordance with experimental measurements.