In reliability assessment, a performance function can typically be defined by the difference between a computed value, for instance, the maximum stress in a structure, and an allowed value. When the performance function includes uncertainties, the problem of accurately and effectively computing the failure risk arises. The uncertainties can generally be of two types: aleatory and epistemic. Aleatory uncertainty is due to inherent randomness while epistemic uncertainty is an uncertainty in the model of a problem.

At the Reliability Modelling and Optimization (RMO) Lab, we develop novel risk/reliability assessment methods using adaptive surrogate modelling techniques as well as space-time varying uncertainty quantification (UQ) and propagation methods. Our ongoing data-driven modelling efforts handles different types of uncertainties including "known unknowns" and "unknown uknowns".

The inherent stochastic nature of loads and structural response poses significant challenges in structural design, especially when high reliability requirements are essential. The inability to properly account for statistical uncertainties may lead to over-dimensioned structures due to the application of unnecessarily large safety-factors. This, in turn, is a major disadvantage from both cost and sustainability aspects.

To address this challenge, we develop novel efficient probabilistic design frameworks that accounts for uncertainties in loads, material as well as dimensional tolerances. An important aspect in our ongoing work is to validate the probabilistic framework against Design Standards and to ensure that the design satisfies target failure probability criteria.  To this aim, a global safety factor is computed based on the proposed probabilistic framework, which is used to validate the partial safety factor method employed in typical Design Standards.

Numerous engineering components experience cyclic loading, which can lead to progressive material degradation and eventual fatigue failure. Ensuring a robust fatigue design is crucial in preventing unexpected breakdowns, thus enhancing the safety and longevity of structures. In our research, we employ and develop fatigue models for both High Cycle Fatigue (HCF) and Low Cycle Fatigue (LCF). In particular, we focus on modelling and mitigating the impact of variability in fatigue life, in order to reduce the uncertainty in predicting component lifespan.