Session: 07-13-01 Optimization, Uncertainty and Probability I

Paper Number: 73306

Start Time: Tuesday, 04:10 PM

73306 - Structural Reliability Estimation of Steel Hall Exhibiting Random Mechanical Parameters 

The aim of this work is to analyse the random response of steel hall under wind excitation located in central Poland. The main frames of this structure have been designed as tapered plate girders made of steel S355W and exhibiting Young modulus expectation equal to 210GPa. Main frames have been equally spaced by distance equal to 7.60m. Gable frames have been spaced from main frames by distance equal to 4.80m and they have been designed as hot-rolled profiles namely square hollow sections (SHS). The Finite Element Method (FEM) analysis has been performed in the Autodesk Robot Structural Analysis 2021 environment (ARSA), where dynamic analysis has been completed using Hilber-Hughes-Taylor algorithm for integrating equations of motion. Numerical solution has been obtained using non-linear dynamic analysis procedure based on Broyden-Fletcher-Goldfarb-Shanno (BFGS) algorithm. Iterative solution has been found with stiffness matrix updating after every iteration. The FEM model consist of 164 beam elements described by linear shape functions and 6 degrees of freedom at each node. Dead load and environmental loads have been introduced according to Eurocodes guidelines respectively, while dynamic excitation of the wind has been assumed as 10 minutes long using some previous experimental measurements.  The results of such analysis have been performed by 0.10s time interval and furtherly saved every 1.0 s of computations resulting in 600 discrete results of each desired state variable. State variables of the greatest interest have been assumed as horizontal displacement of top nodes of the chosen main frame column and also vertical displacement of ridge node of main plate girder. These output data of structural response have been furtherly used in probabilistic analysis, where it has been assumed that Young modulus of the structure is the input Gaussian random variable. The series of computational experiments have been carried out entirely in the ARSA environment for various realizations of Young modulus of the structure to approximate polynomial Structural Response Functions (SRF). This SRF numerical recovery has been performed using the Weighted Least Squares Method and triangular weights distribution. The Iterative Stochastic Perturbation Technique has been contrasted with the Monte-Carlo simulation and semi-analytical method in order to investigate uncertain response of this structure. Expectations of some state variables as well as their higher order probabilistic characteristics have been investigated. Finally, reliability index estimation has been performed and further conclusions have been drawn with respect to reliability index recommendations found in Eurocodes guidelines for basis of structural design.

Presenting Author: Rafał Bredow Łódź University of Technology

Authors:

Rafał Bredow Łódź University of Technology
Marcin Kamiński Łódź University of Technology