Determination of member stiffness on elements jointed by bolts using finite element method and neural network system and bolt design

n DETERMINATION OF MEMBER STIFFNESS OF ELEMENTS JOINTED BY BOLTS USING FINITE ELEMENT METHODS AND NEURAL NETWORK SYSTEM AND BOLT DESIGN (Ph.D Thesis) Murat Tolga ÖZKAN GAZI UNIVERSITY INSTUTE OF SCD2NCE AND TECHNOLOGY July 2003 ABSTRACT Bolted joints have been modelled in 3D, using Finite Element Method (FEM) in ANSYS program according to design parameters namely bolt diameter, screw hole diameter, bolt head diameter and element thickness in order to determine member stiffness in bolted joints. In these models, member stiffness was determined for various materials,thicknesses, modulus of elasticity, Poisson ratios and hole diameters ranging from 3 mm to 30 mm. The data obtained from FEM were applied to Neural Network System so that it gives the conical frustrum angle to determine member stiffness for various dimensions and materials. The results obtained were compared to the literature. A design program for bolted joints, carrying axial loading and shear loading status was prepared in Visual Basic program. The program provides bolted joint designs receiving stress concentration factor from Neural Network System. Bilim Kodu : 708 Anahtar Kelimeler : Member Stiffness, Finite Element Method, Bolted Joint, Neural Network System Sayfa Adedi : 132 Tez Yöneticisi :Prof. Kürşad DÜNDAR Prof. Mahmut GÜLESİN