Journal of Multiscale Modelling

The impact localization in composite panels is assessed using two machine learning techniques: least square support vector machines (LSSVM) and artificial neural networks (ANN) with local strain signals from piezoelectric sensors. Sensor signals from impact experiments on a composite plate as well as signals simulated by a finite element model are used to train and test models. A comparative study shows that LSSVM achieves better accuracy than ANN on identifying location of impacts for a combination of large mass impact and small mass impact, in particular when less data is available for training which is more appropriate for real aeronautical application. Additionally, LSSVM is more capable of identifying new impact events which have not been considered in the training process.

Ceramic foams made by replica techniques containing sharp-edged cavities, which are potential crack initiators and therefore have to be analyzed using fracture mechanical methods. The ceramic foams made of novel carbon bonded alumina are used as filters in metal melt filtration applications, where the filters are exposed to a thermal shock. During the casting process the filters experience a complex thermo-mechanical loading, which is difficult to measure. Modern numerical methods allow the simulation of such complex processes.

As a simplified foam structure an open Kelvin cell is used as a representative volume element. A three-dimensional finite element model containing realistic sharp-edged cavities and three-dimensional sub-models along these sharp edges are used to compute the transient temperature, stress and strain fields at the Kelvin foam. The sharp edges are evaluated using fracture mechanical methods like the $J$-integral technique.

The results of this study describe the influence of the pore size, relative density of the ceramic foam, the heat transfer and selected material parameters on the fracture mechanical behaviour.

Railway axles are subjected to cyclic loading which can lead to fatigue failure. For safe operation of railway axles a damage tolerance approach taking into account a possible defect on railway axle surface is often required. The contribution deals with an estimation of residual fatigue lifetime of railway axle with initial inclined surface crack. 3D numerical model of inclined semi-elliptical surface crack in railway axle was developed and its curved propagation through the axle was simulated by finite element method. Presence of press-fitted wheel in the vicinity of initial crack was taken into account. A typical loading spectrum of railway axle was considered and residual fatigue lifetime was estimated by NASGRO approach. Material properties of typical axle steel EA4T were considered in numerical calculations and lifetime estimation.

Poly(hydroxyl butyrate-co-valerate) (PHBV) is a biopolymer synthesized by microorganisms that is fully biodegradable with improved thermal and tensile properties with respect to some commodity plastics. However, it presents an intrinsic brittleness that limits its potential application in replacing plastics in packaging applications. Films made of blends of PHBV with different contents of thermoplastic polyurethane (TPU) were prepared by single screw extruder and their fracture toughness behavior was assessed by means of the essential work of fracture (EWF) Method. As the crack propagation was not always stable, a partition method has been used to compare all formulations and to relate results with the morphology of the blends. Indeed, fully characterization of the different PHBV/TPU blends showed that PHBV was incompatible with TPU. The blends showed an improvement of the toughness fracture, finding a maximum with intermediate TPU contents.