Abstract

In the last few years, there has been a growing interest in roof structural systems because of their wide
applications in the structure of low-rise buildings. Former research indicates that the analysis of such systems plays a key
role in the design of fasteners with higher performance against the likely damage and failure that occurred due to high wind
events, especially at the connecting point between the fastener and purlin sheet. To find out a solution for improving the
strength of the cladding fasteners, this study aims to present the three methods of geometry optimization for a cladding
fastener that is made of steel austenitic 316. In the first method, two geometry parameters, namely the thread depth and the
thread angle are chosen as the design parameters and manipulated by a genetic algorithm to minimize the maximum von
Mises stress against applied tensile force. The same parameters are swept in the second method in order to generate a set
of data for training a neural network and then the optimum values of the thread depth and angle will be predicted by the
network versus the desired value of von Mises stress as an input. In the third method by proposing the mathematical model,
the optimum geometry values of the fastener will be obtained by maximizing the pull-out force capacity. Finally, functional
tensile tests and the results of the simulation will compare the pull-out force of fasteners designed by three methods.