Abstract

Optimization of stacking sequence would enable the designer to efficiently exploit tailoring abilities of such filament wound cylinders. Hence, the main aim of the paper was to perform optimization of stacking sequence using a genetic algorithm for two representative models with the first model having a length of 1650 mm and diameter 350 mm and second model with length 238 mm and diameter 184 mm. MATLAB programs for calculating critical buckling pressure (CBP), stiffness matrix and fitness evaluation function were developed using Sander’s type analytical model. Then, the parametric study was performed to investigate the effect of geometrical dimensions on CBP, determine optimum settings for MATLAB Genetic Algorithm (GA) toolbox and a number of elements for performing Eigen buckling analysis in ANSYS. Considering the optimal parameters, optimization of stacking sequence using MATLAB GA tool for increasing depth of operation of both the models and decreasing weight of the first representative model was performed. The parametric study indicated that CBP is inversely proportional to (5/2)th power of D/t ratio and first power of L/D ratio. The CBP was maximized for a stacking sequence that leads to the dimensionless bending stiffness ratio nearly equal to 0.1. Thicknesses of the full-scale models of first representative cylinders were 13.2 mm and 18 mm of carbon/epoxy and glass/epoxy respectively for a depth of operation of 1000 m. CBP of the second model with a thickness of 5 mm was 16 MPa and 7.8 MPa for carbon/epoxy and glass/epoxy respectively. For the first model, substantial improvements of over 85 % and 44 % in depth of operation for carbon/epoxy and glass/epoxy were achieved, weight reductions of 13 % to 23.4 % for an operating depth of 1000 m were achieved. For the second representative model, improvements of 52.36 % and 19 % in depth of operation for carbon/epoxy and glass/epoxy respectively were achieved. A deviation of 5 – 25 % and 10 – 40 % was observed for glass/epoxy and carbon/epoxy respectively when compared with results obtained from Eigen buckling analysis using ANSYS.

Authors and Affiliations