Analysis of the antiplane shear fracture of elastically coupled composite laminates by using integration of the acoustic emission technique and machine learning algorithms

SONATA 21

Funding Organization: National Science Centre

Project title: Analysis of the antiplane shear fracture of elastically coupled composite laminates by using integration of the acoustic emission technique and machine learning algorithms

Agreement number: UMO-2025/59/D/ST11/00238

Project implementation period: 21.08.2026 r. - 20.08.2029 r.

Principal Investigator: dr inż. Jakub Rzeczkowski

Project value: 1 339 560,00 PLN

Funds granted for Lublin University of Technology: 1 339 560,00 PLN

Abstract: The main scientific goal of the project is experimental and numerical study on influence of elastic couplings phenomena and boundary conditions on delamination process in multidirectional laminates subjected to the anti-plane shear loading. Results obtained in the project will deliver a new knowledge about relationship between presence of the elastic couplings and various failure mechanisms occurring in composite laminates under the transverse shear loading conditions, as well as they will allow to develop an experimental test setup together with the intelligent solutions to detection of delamination initiation and automatic classification of various failure modes. The need for better recognition of the mode III delamination phenomenon, leading to determination of the respective fracture toughness parameter GIIIC (critical strain energy release rate, c-SERR) became a serious problem in recent years, along with the growing application of the FRP composites in the aircraft and other industries, the more no commonly accepted standard exists. The popular edgecrack torsion test requires specialized beam-like specimens. The alternative can be the split cantilever beam and similar procedures, enabling adaptation of the simple specimens, commonly used in mode I and mode II tests. This approach could reduce the total cost of experimental tests and decrease the total time of determining the SERR values of the three fracture modes for a given laminate – all necessary in finite element simulations. In the ECT test, the mode III cracking can be contaminated with mode I and II fracture. As shown recently by the Principal Investigator (PI) and others, also in the SCB configuration fake modes I or II may occur, especially at the beam edges. This demands improvements in existing test methods. Within the framework of the project two main groups of multidirectional laminates will be examined: fully uncoupled and elastically coupled; within the latter two kinds of coupling will be considered: the bending-extension and the bending-twisting. The FE numerical analyses will be conducted in the Abaqus/CAE by using the virtual crack closure technique and the cohesive zone model (CZM) in the SCB (special testing clamp patented, Patent No. P436089) and ECT type configurations, at various boundary conditions (BC). All experiments will be supported with the acoustic emission technique. Registered AE signals will be process to extract typical waveform features (the frequency signature will be determined by using the Hilbert Huang mode decomposition) and transform by using the wavelet transformation to present the signals in the form of spectrograms. The AE features will be clustered by using the machine learning k-means technique to create typical damage signatures library. For the latter, the collection of the AE frequency-time representations will create an input data set to deep neural network that will be trained for classification purposes. In addition, a posterior fractographic observations of real fracture surfaces after experimental tests will be conducted. The novelty of the research deliberated for the project is development of the SCB-type test procedure utilizing synergy of the AE technique, application of the Hilbert Huang transform (elaborated by NASA in recent years) and the AI algorithms to automatic detection of delamination initiation and classification in real time of typical failure mechanisms occurring inside tested laminate. The results of the project will broaden the knowledge on delamination processes under the antiplane shear loading and they can contribute to development of novel research field of intelligent composite materials testing methods.