Abstract: This dataset of scans was used in a publication that deals with the computational determination of fiber volume contents from µCT scans. Specimens from two carbon fiber reinforced polyamide 6 plaques are used that were produced in the long fiber reinforced thermoplastic direct (LFT-D) process. From the first plaque, three specimens each from both the charge and flow area of the plaque are included with dimensions of 10 mm $\times$ 10 mm $\times$ 3 mm, 20 mm $\times$ 20 mm $\times$ 3 mm and 30 mm $\times$ 30 mm $\times$ 3 mm, respectively. From the second plaque nine specimens were extracted in a regular 3 $\times$ 3 grid all with the dimensions 25 mm $\times$ 25 mm $\times$ 3 mm. Exact cutting plans can be taken from the publication. The specimens are cut out of the plaques through waterjet cutting. The material constituents, with which the final CF-PA6-LFT-D plaques were manufactured, are Zoltek PX35 commercial carbon fibers and a PA6 by the company DOMO. The scans of the first plaque were done with a Zeiss Metrotom 800 (cone-beam) μCT system at wbk Institute of Production Science at KIT with the flat panel detector PaxScan2520V with 1536 pixel $\times$ 1920 pixel. The samples of the second plaque were scanned in an YXLON-CT (Yxlon International CT GmbH, Hattingen, Germany) precision μCT system at Institute for Applied Materials (IAM-WK) at KIT with a μ-focus X-ray transmission tube with tungsten target and a PerkinElmer (Waltham, MA, USA) Y.XRD1620 flat-panel 2048 pixel $\times$ 2048 pixel detector. More details can be found in the corresponding publication.
TechnicalRemarks: The 15 scans are available as .raw- and .mhd-files, which can be read in Python with, e.g., the SimpleITK package, or in ImageJ (FIJI). The entirety of the scans are available twice: One set only has small brightness and contrast adjustments and cutting at the edges, but is otherwise in the original 16 bit data format. This set was used for the training of the convolutional neural network. The other set was cut more and is converted into 8 bit, which was used for the thresholding algorithm.
