Functionality, quality and efficiency of engineered systems, products and processes more than ever depend on trustworthiness of “computer-aided” analytical tools for design and operation. Trustworthiness of such mathematical analytics and prediction is affected by several sources of uncertainty, such as limited data availability or accuracy, approximations in modeling, and computational processing. Uncertainty quantification (UQ) systematically addresses these still too frequently neglected issues. UQ research is expanding and already has made huge progress in theory and application impact to quantify, and possibly reduce, uncertainty in the predictions of system behavior. Areas of application include climate and environmental modeling, energy generation, control and manufacturing, and process and system design. Uncertainty quantification has moved beyond Monte Carlo sampling of inputs to adaptive design of computer experiments, use of surrogate models such as Gaussian process models, development of stochastic expansion approximations, calibration (including uncertainty both in experimental data and in model form), model validation, the quantification of prediction uncertainty, and the development of UQ algorithms for multi-scale and multi-physics applications.
The workshop will bring together industrial scientists, lab scientists, and university based researchers to share UQ state-of-the-art application experience, best practices, and future challenges in diverse applications and from different sectors, including aerospace and automotive applications, engine design, energy (nuclear, wind, solar, etc.) and global climate change. Presentations on academic UQ research and progress will open opportunities to transfer results for application problem solutions. Special focus will be given to exploit the joint potential of data-driven statistical approaches and model-based methodology. Moderated discussion sessions will catalyze joint research activities and knowledge transfers.
* A number of 45-minute talks with 15 minutes for discussion following each.
* One poster session.
* 3 one-hour-long moderated discussion sessions