A probabilistic FEM approach for the structural design of glass components

A new probabilistic computational methodology aiming for a safer and optimized design of glass components is proposed, overcoming the drawbacks of the currently employed design approaches.The methodology, which adopts a stress intensity factor-based fracture criterion, can be applied to predict the load bearing capacity associated to any given probability of failure of elements having arbitrary geometry, support conditions and edge flaws scenario.The main novelty consists in the use of the extended finite element method for the numerical modelling of the structural elements, taking advantage of its intrinsic capability to deal with multiple cracks without adapting the mesh topology and the possibility to directly evaluate the stress intensity factor at the tip of the cracks without any post-processing procedure. Besides, because of the stochastic nature of the problem, where the flaws size is the random variable, the Monte Carlo method is used to obtain the cumulative distribution function of the failure load, from which the load bearing capacity is derived.Several case studies are reported to demonstrate the accuracy and reliability of the method. It is also shown that, depending on the stress gradient along the glass component, the developed method provides load carrying capacities larger than the predictions of a stress-based approach, by an extent variable between 21% and 83%.

Automated summary

This study proposes a new probabilistic computational methodology for the safer and optimized design of glass components, which overcomes the limitations of current design approaches. The methodology uses a stress intensity factor-based fracture criterion to predict the load bearing capacity of elements with arbitrary geometry, support conditions, and edge flaws. By utilizing the extended finite element method, the method can handle multiple cracks without changing the mesh topology and directly evaluate the stress intensity factor. The Monte Carlo method is employed to obtain the cumulative distribution function of the failure load due to the stochastic nature of the problem. The method is demonstrated to be accurate and reliable through several case studies and provides larger load carrying capacities than stress-based approaches, ranging from 21% to 83%, depending on the stress gradient along the glass component.