Effect of cutout on stochastic natural frequency of composite curved panels

Abstract

The present computational study investigates on stochastic natural frequency analyses of laminated composite curved panels with cutout based on support vector regression (SVR) model. The SVR based uncertainty quantification (UQ) algorithm in conjunction with Latin hypercube sampling is developed to achieve computational efficiency. The convergence of the present algorithm for laminated composite curved panels with cutout is validated with original finite element (FE) analysis along with traditional Monte Carlo simulation (MCS). The variations of input parameters (both individual and combined cases) are studied to portray their relative effect on the output quantity of interest. The performance of the SVR based uncertainty quantification is found to be satisfactory in the domain of input variables in dealing low and high dimensional spaces. The layer-wise variability of geometric and material properties are included considering the effect of twist angle, cutout sizes and geometries (such as cylindrical, spherical, hyperbolic paraboloid and plate). The sensitivities of input parameters in terms of coefficient of variation are enumerated to project the relative importance of different random inputs on natural frequencies. Subsequently, the noise induced effects on SVR based computational algorithm are presented to map the inevitable variability in practical field of applications.

Frequency domain homogenization for the viscoelastic properties of spatially correlated quasi-periodic lattices

“…comb core in a sandwich structure may increase the degrees of freedom for the entire system up to such an extent that can make the overall process unmanageably complex and prohibitively expensive for simulation. In case of uncertainty quantification using a Monte Carlo based approach, the problem aggravates as large number of expensive finite element simulations are needed to be carried out (Dey et al, 2017(Dey et al, , 2016aHurtado and Barbat, 1998;Mahata et al, 2016;Mukhopadhyay, 2017;Mukhopadhyay et al, 2015. Application of surrogate based approaches to achieve computational efficiency, as adopted in many of these papers, does not make the analysis physically insightful and this approach often suffer from lack of confidence in the predicted results.…”

Probing the shear modulus of two-dimensional multiplanar nanostructures and heterostructures

“…Such development can help to bring about the much-needed impetus in the research of two-dimensional materials, which is often hindered due to the need for carrying out computationally expensive and time consuming simulations/ laboratory experiments and availability of interatomic potentials. Besides deterministic analysis of shear moduli, as presented in this paper, the efficient closed-form formulae could be an attractive option for carrying out uncertainty analysis [99][100][101][102][103][104] following a Monte Carlo simulation based approach.…”

Stochastic impact responses analysis of functionally graded plates

“…This drawback of MCS has been mitigated by developing surrogate modelling approach [45][46][47][48]. Stochastic natural frequency of noise-induced support vector regression (SVR) model for laminated composite curved panels is analysed by Dey et al [49]. The effect of noise in surrogate model is quantified by Mukhopadhyay et al [50], whereas Dey et al [44] presented the effect of noise using polynomial neural network approach in uncertainty quantification of natural frequency, and Chakraborty [51] predicted delamination in laminated composite using ANN approach.…”