Dinamik yükler altında bağlantı elemanlarının davranışlarının incelenmesi

This work aims to improve the understanding of bolt behaviors in dynamic load scenarios, particularly those influenced by transverse loads. In order to forecast fastener outcomes under varying load situations, a comprehensive investigation was conducted encompassing experimental, theoretical, machine learning, and simulation-based study methodologies. Findings indicates that bolts subjected to vibratory loosening are more susceptible to experiencing increased friction in their bearing and thread sections. This is primarily caused by wear in the contacting regions resulting from both macro and micro slips. Therefore, the capacity to reuse fasteners becomes uncertain in certain situations. Experimental results have shown that the type of coating used also has a significant effect on the behaviors observed. A simulation model was developed to address the challenges of conducting repeated experimental testing, specifically focusing on the factors of tightening, friction change, and loosening. The simulation approach enables researchers to predict the effects of fastener loosening risks. The solve time for the loosening simulation might be significantly increased depending on the thread mesh density. A neural network-based deep learning technique and model were created to address this problem. Model was trained using both experimental data and simulated outcomes. To manage the function errors, the Mean Squared Error (MSE), Mean Squared Deviation (MSA), and R-squared values are used. The model loop built in this context is designed to optimize the parameters during the training process and identify the most favorable scenarios. This enables us to input experimental and simulation data into the neural network in the future, hence enhancing the program's dependability without making any modifications to the underlying code. The last part of the study primarily examines the theoretical foundation of bolt stresses and the lifespan of loosening effected bolts. Calculations were conducted using cantilever beam theories, which is a widely used method for analyzing load instances in bolted joints. A methodology has been built to forecast the fatigue life of the bolt based on its transversal fatigue life when subjected to repetitive tightening. This methodology utilizes cumulative damage theories. Results obtained from the experimental investigation and theoretical assumptions fall within an acceptable margin of error. This theory allowed for the transformation of transversal displacement inputs into lateral shear and force values. These values were then used to make predictions about the lifespan of structures under loosening conditions. This approach also incorporates the computation of bearing friction affects and load transfer capabilities. An increase in the shear load capacity results in an increase in the bending moment while maintaining the same transverse displacement. This increase in bending moment leads to a decrease in the fatigue life of the bolt. This is also consistent with the findings of the friction change test conducted during repeated tightening Junker vibration testing, where samples fail due to bending fatigue. It is crucial to consider the reusability of the fastener when analyzing high transverse loads so as to improve the precision of its estimated lifespan.