GCRIS

Now showing 1 - 2 of 2

Material Model Calibration of Fiber Reinforced Concrete Using Deep Neural Network
(01. Izmir Institute of Technology, 2023) Yaşayanlar, Yonca; Saatcı, Selçuk; Erdem, Tahir Kemal; Saatcı, Selçuk; Erdem, Tahir Kemal
The numerical modeling of fiber reinforced concrete (FRC) structures is quite challenging due to the material's heterogeneous and anisotropic nature. The majority of material models that are suitable for regular concrete are not able to account for the FRC material's increased tensile capacity and ductility. In this study, a calibration method is proposed that is simple and effective for modeling FRC structures using a selected concrete material model. The Karagozian and Case (K&C) material model in LS-DYNA is capable of representing the ductile nature of FRC, and its parameters related to tensile behavior were calibrated to reflect the tensile-softening behavior. The calibration process was executed using the uniaxial direct tension test results of two different FRC mixtures. In addition, single element numerical models were constructed using LS-DYNA under uniaxial tension. The tensile parameters of K&C were varied over a wide range using single elements to form a database. Then, a Deep Neural Network (DNN) was constructed to pass the database through and find the K&C parameters that best matched the experimental uniaxial test results. The proposed methodology was tested under static and high-strain rate loading conditions, and the results were compared to the experimental findings. The performance of the DNN-achieved parameters was found to be satisfactory. The results showed that the DNN-calibrated parameters were able to accurately predict the behavior of FRC structures under static and dynamic loading conditions.
Advanced Material Characterization and Modeling the Foreign Body Impact Damage Initiation and Progression of a Laminated Carbon Composite
(01. Izmir Institute of Technology, 2023) Bayhan, Mesut; Taşdemirci, Alper; Güden, Mustafa
The coupon level composite sample tests and the accompanying numerical models were carried out to predict the response of woven carbon fiber composite structures against impact. The numerical models of the coupon-level tests were implemented in LSDYNA software using the MAT_162 and MAT_58 composite material models. The results obtained by both quasi-static and dynamic tests were used to determine their constants. In addition to the tests that were used for the determination and calibration of the material model parameters, separate tests and their models were performed for the validation, including punch shear tests and low-velocity impact tests. It could be said that the material models examined were considered comprehensive and precise as the experimental results were well predicted by the numerical models. Also, the rate sensitivity of the woven carbon composite in the in-plane and thickness directions was investigated experimentally and numerically. In the tests, the DIC method was employed in the determination of the displacement and strain of the specimen. Based on the results obtained, it was concluded that the in-plane tensile properties are rate insensitive. Besides, the simulations of the component level tests, such as bird strike and drone impact, were established to investigate the damage initiation and propagation within the composite. It was found that the drone impact results in more severe damage compared to the bird impact. It is worth noting that the development of such precise composite material models to simulate dynamic loadings will definitely shorten the time between the beginning of designing and the component testing.

Phd Degree / Doktora

Browse

Filters

Settings

Sort By

Results per page

Search Results