Reliability in the characterization of fiber length distributions of injection molded long carbon fiber composites

Abstract

The mechanical properties of a discontinuous fiber reinforced composite material are dependent on the length of the reinforcing fibers. Long-fiber thermoplastic (LFT) composites offer a compromise between ease of manufacture and mechanical properties for structural applications and are conventionally made using fibers that are initially 10-14 mm long. However, fiber attrition during the injection molding process significantly reduces the final fiber lengths in molded composite and produces a fiber length distribution (FLD), which depends on the processing parameters. Thus, accurate FLD measurements are important for reliable predictions of the composite properties. Here, we present a detailed methodology for experimentally measuring the FLD in LFT composites based on best practices. Possible biases occurring during the measurement process are discussed and their effect on the measured FLD and averaged fiber length values are evaluated. We then employ a Mori-Tanaka implementation of the Eshelby model to predict the elastic stiffness of the LFT composite and analyze the sensitivity of the stiffness values to the biases in the FLD measurements. POLYM. COMPOS., 00:000-000,

Predicting Microstructural Void Nucleation in Discontinuous Fiber Composites through Coupled in-situ X-ray Tomography Experiments and Simulations

“…The experimentally-based DVC analysis showed that the highest strain values were localized in the region of microvoid nucleation, which corresponded spatially to the hydrostatic stress hot-spot predicted by the finite element model. Typically, engineers are not interested in fragments of fibers with aspect ratios close to one because they contribute almost no load bearing capability when compared to their longer cylindrical neighbors present in the microstructure [28][29][30] . Therefore, techniques and algorithms to detect and characterize fibers are typically not suitable for ≤ l d / 1, which can be easily confused with noise 30,31 .…”

“…Typically, engineers are not interested in fragments of fibers with aspect ratios close to one because they contribute almost no load bearing capability when compared to their longer cylindrical neighbors present in the microstructure [28][29][30] . Therefore, techniques and algorithms to detect and characterize fibers are typically not suitable for ≤ l d / 1, which can be easily confused with noise 30,31 . In this work, the fragment of glass was manually observed in the in-situ tomography images, detected as a fragment of a fiber, included in the microstructural simulation, and proved to be a region of high hydrostatic stress and a region of corresponding microvoid nucleation, as was seen in Fig.…”

Characterization and Model Validation for Large Format Chopped Fiber, Foamed, Composite Structures Made from Recycled Olefin Based Polymers

“…Of interest, then, is the resulting fiber breakage caused by the high shears during the manufacturing processes. Thus, the as manufactured aspect ratio is of interest and an approach similar to that of [16] will be performed. A Q50 Thermogravimetric Analyzer (TA Instruments, New Castle, DE, USA) was used to observe the decomposition of GFPP in the presence of air and nitrogen, and typical results are shown in Figure 4.…”

“…The reduction in the fiber length distribution depends on process parameters such as back pressure, speed of revolution, barrel temperature and the polymer melt shear rates [15]. In quantifying the fiber length distribution, it was shown in [16] that over 800 fiber samples, properly selected from within a sample set, are required to achieve an error less than 5%. Fiber measurements can be biased based on improper sampling on the edges of the sample area, fiber breakage during sample preparation, and improper discrimination for very short and very long fibers, the details of which are expounded upon in Section 2.2.…”

“…Fiber measurements can be biased based on improper sampling on the edges of the sample area, fiber breakage during sample preparation, and improper discrimination for very short and very long fibers, the details of which are expounded upon in Section 2.2. The weight-based aspect ratio gives preference to the contribution of the longer fibers and is more appropriate for use in micromechanical predictions, whereas the number based aspect ratio gives equal weight to fully intact fibers and the multiple sub-structures of a fiber that is broken during manufacturing [16].…”