Uncertainty quantification in engineering

Abstract

Computer simulation has become an important tool in engineering sciences in the last two decades. Nowadays computer models help assess and optimize civil or mechanical structures and systems, from the early design to the manufacturing or construction phase. All along their life time, systems are prone to uncertainties: natural variability of their characteristics (e.g. dimensions, physical properties) in case of mass production, uncertainties in the environment (e.g. natural hazards and climatic loads) for infrastructures. This has lead engineers and researchers to introduce statistics and probability tools to address uncertainty challenges in conjunction with the computer models that reproduce the physical behaviour of the system.

So called uncertainty quantification techniques aim at identifying the sources of uncertainty affecting the system and quantify their impact onto the system performance for the sake of robustness, risk analysis (probability of system failure and associated consequences) and sensitivity analysis (detection of the “important” uncertain parameters). The lecture will introduce the basics of uncertainty quantification from the engineering perspective with various application examples.