Single-Objective Optimization of Grooved Thin-Walled Circular Tubes in Axial Loading

In this paper, single objective optimization of grooved thin-walled circular tubes in axial loading is performed. An explicit finite element model of a grooved tube under axial load is generated using LS-DYNA software. The output data of the FE model is validated by means of comparison between corresponding results and similar experimental tests. A MATLAB code is generated which is able to generate the LS-DYNA inputfile, run LS-DYNA solver and elicit the results for a tube with arbitrary mechanical and geometrical properties. The code is also able to generate and run multiple FE models with a matrix of desired input data. This code is then used in order for generating and solving required number of FE models with various properties, as the input data for two neural networks. The networks are means to predict the amount of absorbed energy and the peak load during the loading process, respectively. A mathematical function is then defined as an index for combination of the absorbedenergy and the peak load, which varies ascending with the peak load and descending with the absorbed energy. The final objective is to minimize this function to reach the best combination of properties which yields the best combination of energy absorption and maximum load. To minimize this function, single objective genetic algorithm is used and this function is considered as the fitness function of the algorithm. The result of minimization yields the best geometrical properties of a tube with fixed mechanical (and some geometrical) properties.

Keywords:

optimization , grooved tube , geneticalgorithm , neural network , axial loading