Open AccessDESIGN OPTIMIZATION OF A CIVIL STRUCTURE USING NUMERICAL SIMULATIONS
Author(s) -
Florian Vlădulescu
Publication year - 2020
Publication title -
mechanical testing and diagnosis
Language(s) - English
Resource type - Journals
eISSN - 2668-571X
pISSN - 2247-9635
DOI - 10.35219/mtd.2020.2.01
Subject(s) - parametric statistics , finite element method , sensitivity (control systems) , design of experiments , process (computing) , optimal design , tower , computer science , shell (structure) , parametric model , integer (computer science) , mathematical optimization , matrix (chemical analysis) , optimization problem , variable (mathematics) , mathematics , algorithm , engineering , structural engineering , statistics , mechanical engineering , electronic engineering , mathematical analysis , materials science , composite material , programming language , operating system
This optimization study aims to determine the effect of each input parameter on the output parameters, how the input parameters can interact with each other and also it is emphasized the determination of the values for the input parameters that optimize the responses. In this study, the objective is to obtain an optimal configuration for a resistance structure specific to a telecommunications tower. For this purpose, a variable geometric model using design parameters is created, based on which a 3D finite element model (FEM) is obtained, which is used in the optimization study. The FE model is updated automatically for each version of geometric model and is made using beam and shell elements. Design of Experiments (DOE) methodology allows for using a mathematical model that predicts how input parameters interact to create output responses in an optimization process. Parameters’ correlation and monitoring allow for identifying important parameters and the correlation matrix and sensitivity graphs also help understanding the parametric relationships. Variation limits for design parameters are defined and these parameters can have integer or fractional values.