Recycling of Thermoplastic Unidirectional Tapes: An Image‐Analysis Method for Measuring Residual Glass‐Fiber Length in Convoluted Fiber Clusters

ABSTRACT Endless fiber‐reinforced thermoplastic composites face unresolved recycling challenges, with incineration remaining the dominant end‐of‐life option. High‐quality mechanical recycling requires converting them into short‐fiber reinforced materials with well‐defined fiber length distributions. This necessitates a reliable, fast, and low‐cost fiber length measurement method. This study develops and validates an offline image‐based technique for determining fiber length distributions, including dense or convoluted fiber clusters. Shredded unidirectional (UD) glass fiber‐reinforced polypropylene (PP) tapes were used, with the matrix removed by calcination. Fiber bundles were imaged using a Keyence VHX‐7000 digital microscope or a flatbed scanner. Two sampling methods were evaluated to optimize fiber dispersion and minimize overlap. The rule‐based image analysis algorithm distinguishes isolated fibers from convoluted fiber clusters by contour geometry. Clusters are skeletonized and their junctions (X‐, T‐, Y‐, and L‐shaped branch points) are located with morphological hit‐or‐miss kernels, cutting the cluster into fragments. A directional search‐cone algorithm, defined by a maximum reach and lateral deviation from the fragment axis, iteratively reconnects the fragments into individual fibers. This enables fiber length determination in dense clusters without training data or manual intervention beyond an initial parameter setting. Compared to digital microscopy, flatbed scanning with the software reduced measurement time by two‐thirds. Validation on synthetic images with known ground truth ( r = 0.976, 90% of fibers within ±10% of true length), on real fiber clusters and against commercial systems (Anton Paar, FASEP), confirmed high accuracy and robustness. The method offers a time‐efficient, robust tool supporting the circular use of fiber‐reinforced composites.