The Journal of Strain Analysis for Engineering Design

Analysis of stresses in components of even modestly complex geometries often require the use of finite element analysis (FEA). For testing a large number of design options quickly, FEA can be time consuming and provides more accuracy than required. In this project, a machine learning-based system is developed to provide quick and approximate solutions to stress analysis problems in parametrised compressor disc geometries. Simple mechanics problems were completed preliminarily to test the practicality of machine learning approaches for this application. This included applying instance selection by the k -medoids algorithm to a 2D FEA problem. A parametrised compressor disc geometry was designed and defined by eight dimensions. Stress fields were produced from two superimposed loading schemes: the rotational body force and the force exerted on the disc by the blades 495,338 samples of training data were collected from 4374 FEA simulations. Four networks were trained to predict stresses caused by each loading scheme in order to produce stress fields. The best network was of the structure 9/20/20/2 and used normalised training data. For the geometry tested, it predicted stresses with a root-mean-square error of 1.51%. The code took 0.7 s to run in total, from start-up to completion of a stress plot. The importance of the inputs in the training data set were scored with a feature selection algorithm to aid further optimisation of the system. The low computation time makes this system suitable for the early stages in a design process.

In this article, a model of two dimensional problem of generalized thermoelasticity for a homogeneous isotropic nonlocal microstretch saturated porous medium with variable thermal conductivity is established. The enunciation is applied to generalized thermoelasticity theory based on Moore-Gibson-Thompson heat conduction model. The formulae for amplitude ratios and energy ratios corresponding to various reflected waves have been presented, when a set of coupled waves strikes obliquely at the boundary surface of the assumed model. The numerical values of modulus of reflection coefficients are presented graphically to depict the effects of microstretch parameter, porosity, variable thermal conductivity, and non local parameter. Expressions of energy ratios have also been obtained in explicit form and are shown graphically as functions of angle of incidence. It has been verified that during the reflection phenomenon, the sum of energy ratios is equal to unity at each angle of incidence. Results of some earlier workers have been deduced from the present formulation.

Porous defects in functionally graded materials (FGMs) significantly change their mechanical behavior, in particular, affecting the displacement of the neutral plane and stress distribution. Defects such as porosity reduce material bending strength while enhancing compression strength and can lead to significant transverse effects. The purpose of this study is to clarify the influence of porosity on transverse effects based on new kinematic model. This research consider in detail a particular case of this model, that allows take into account two Poisson`s effects separately. This study examines the influence of porosity on FGM, highlighting the critical role of changes in Poisson’s ratio through thickness. The theory used for analytical solution satisfies the nullity of the shear stresses at the upper and lower surfaces of the plate without using the shear correction factor. The network of pores proposed to be empty or filled with low-pressure air and material properties graded by thickness according to power-law and exponential models. Navier solution method is used to solve the governing differential equations of equilibrium. The concentration of porosity at the center of the plate or in separate layers results in a reduction in normal tensile stresses along the longitudinal coordinate, suggesting that strategic placement of porosity can be used to reduce stress concentrations, potentially enhancing the structural integrity and performance of functionally graded materials under mechanical loading.

Fatigue damage on the surface of brake discs can seriously affect the train braking safety. In this paper, the effect of various braking conditions on the crack propagation of train brake discs is investigated based on the co-simulation of Abaqus and Franc3D. The relationship of the braking mode, initial speed, braking time and braking force to the stress intensity factor, crack propagation rate, and propagation life is discussed. The findings show that the initial speed is the major factor influencing disc crack propagation during emergency and service braking, while the braking force and time are more important during ramp braking. Besides, the temperature increasing rate and temperature field distribution are significant factors influencing crack propagation. Thus, reasonable design of braking condition parameters can help slow down the crack propagation of brake discs and ensure the safety of braking.

This study presents a semi-analytical solution that can calculate the free vibration frequencies of functionally graded nanobeams with three distinct pore distributions under both deformable and rigid boundary conditions, based on nonlocal elasticity and Levinson beam theories. The novelty lies in the incorporation of transverse springs at both ends of porous functionally graded nanobeams and introducing a general eigenvalue problem dependent on the stiffness of these springs. This solution provides vibrational frequencies considering Levinson beam theory, non-local elasticity theory, spring stiffnesses, porosity coefficients, and temperature change. Additionally, the vibrational behavior of these porous nanobeams is explored through machine learning (ML) techniques. Four ML models namely artificial neural network (ANN), support vector regression (SVR), decision tree (DT), and extreme gradient boosting (XGB) are trained to predict the natural frequencies of nanobeams with varying pore distributions. The Sobol quasi-random space-filling method is employed to generate samples by altering input feature combinations for different porous nanobeam distributions. Model performance is evaluated using different performance indicators and visualization tools, with optimal hyperparameters determined via a Bayesian optimization algorithm. Results underscore the efficacy of ML models in predicting natural frequencies, with SVR and ANN demonstrating superior performance compared to XGB and DT. Notably, SVR and ANN exhibit exceptional R² values of 0.999, along with the lowest MAE, MAPE, and RMSE values among the models assessed. At the end of the study, the effects of various parameters on porous gold (Au) nanobeams using the solution of the presented eigenvalue problem are discussed.