PSI - Issue 83

Miloslav Kepka et al. / Procedia Structural Integrity 83 (2026) 138–145

141

The DePriSS project (Development of "3D printing and thermal spraying systems" for applications with dynamic and impact loads) was led by the Research and Testing Institute Plze ň (VZU Pilsen) and involved not only the staff of the RTI research center, but also the Institute of Scientific Instruments of the Czech Academy of Sciences, Fraunhofer IKTS and Opole University of Technology. The DePriSS project focused on combining advanced 3D printing and thermal spraying technologies to develop components with improved wear and fatigue resistance under dynamic and cyclic loads. By integrating these two additive manufacturing methods, the project aimed to overcome traditional manufacturing limitations and achieve superior surface properties for high-tech applications. Within the project duration, several specific achievements were reached: Patent Application: A Czech national patent (No. 309502, 2023) was secured for an innovative method enabling direct thermal spraying on 3D-printed components without requiring grit blasting, reducing environmental impact and increasing manufacturing efficiency. Industrial Demonstrator: Developed and validated a 3D-printed injection mold core with HVOF-applied surface coatings, showcasing improved cooling efficiency and extended operational life. New Material Development: Designed a cobalt-free feedstock powder optimized for dynamic impact resistance, offering sustainable and high-performance alternatives for wear-prone industrial components. Collaborative Innovations: Established methodologies for fatigue testing and dynamic impact analysis of coated systems, ensuring the project's outcomes are applicable across multiple industries such as aerospace, energy, and automotive. High-Quality Publications: Results have been disseminated so far through five peer-reviewed articles and sixteen conference presentations, advancing knowledge in the fields of additive manufacturing and surface engineering. For example, Böhm et al. (2023) presented selected S–N curves for EN 1.2709 steel after printing with the selective laser melting method. The characteristics were compared, and conclusions were presented regarding the resistance of this material to fatigue loading, especially in the tension–compression state. A combined general mean reference and design fatigue curve was presented, which incorporated the author's experimental results as well as those from the literature for the tension–compression loading state. The design curve may be implemented in the finite element method by engineers and scientists in order to calculate the fatigue life. The study presented again by Owsi ń ski et al. (2025) investigated the fatigue resistance of additively manufactured MS1 steel coated with thermally sprayed WC Cr3C2-Ni and Cr3C2-NiCr, applied using High Velocity Oxygen Fuel technology. The research focused on evaluating the effects of these coatings on fatigue life under varying mechanical loading conditions. It was shown that such treatments greatly improve fatigue resistance, especially at low stress levels, by delaying crack initiation and slowing down their propagation. Nevertheless, when stress levels are higher, the benefits from coating diminish and differences in fatigue strength between coated and uncoated specimens are unnoticeable. Under torsional loads, the coatings are rapidly damaged, which may suggest that the driving force behind wear is surface shear stress and that further improvement of the durability of coatings is necessary. 3. Introduction to using machine learning for powder bed quality control Selective laser melting (SLM) is a subtype of Laser powder bed fusion (LPBF) technology. A thin layer of powder is deposited on the substrate, and the laser selectively melts the current cross-section of the part, and again, a new layer of powder is spread. This process repeats until the final part is built. It allows the creation of highly complex geometry that could not be manufactured with any conventional technology. However, this “layer-by layer” process might be a source of defects in the printed part. Therefore, the quality of each layer is essential for ensuring the mechanical properties. In this paper, we describe possibilities of using Convolutional Neural Networks (CNNs) to monitor layer quality. CNNs are a subtype of neural networks designed for effective image processing. The history of CNNs dates to the 80s, when the first models were created. Lately, as architecture became more sophisticated, hardware was still a limitation. The most significant breakthrough came in 2012 when a model called AlexNet was introduced and succeeded in the ImageNet competition. This came with using effective GPU CUDA cores instead of CPUs. These days, CNNs are widely used for computer vision tasks such as image classification, detection, and segmentation. The basic principle is using a convolution layer, which performs edge, texture, shape, and more complex detections. Nowadays, a wide range of CNNs exists for specific applications in the computer vision field.