University of Cambridge > Talks.cam > Engineering Design Centre Seminars  > Uncertainty Quantification in Thermofluids: Key Tools, Applications and Perspectives

Uncertainty Quantification in Thermofluids: Key Tools, Applications and Perspectives

Add to your list(s) Download to your calendar using vCal

If you have a question about this talk, please contact Mari Huhtala.

Uncertainties are ubiquitous throughout the field of applied thermofluids. Thus, developing techniques to identify and rigorously quantify these uncertainties is of significant interest to companies up and down the U.K.

The field of uncertainty quantification has experienced a Renaissance of sorts over the past decade. It has become much more than the science of running deterministic computer simulations at different boundary conditions. It has grown to encompass novel techniques for exploring parameter sensitivities, for finding dimension reducing subspaces, and for gaining insight into the underlying thermofluid mechanics of simulation-driven problems using machine learning approaches. Succinctly stated, its physical insight on a tight budget leading to statistically sound decision making.

In this rather broad talk, I present three different problems that reflect the levels at which uncertainties need to be quantified and the methods used. First, there is the sensor-level, where one tries to convert a voltage change to a pressure or temperature. Uncertainties here are grouped into random and systematic and need to be properly accounted for by understanding the material properties of the sensor, its range, and calibration particulars. Next, there is the model-level, where one typically uses a spatial average of a set of temperatures and pressures as an input to estimate the performance of a component—the output of the model. Methods that use response surfaces are typically adopted here for reducing the computational cost of running the models. Finally, the output of these models (and their uncertainties) inform a system-level exchange rate matrix, where one tries to understand which uncertainties impact whole-system cost and efficiency. Linearized or first-order methods are usually adopted at this level for rapid analysis.

I close this talk by offering a glimpse of the techniques we are currently developing to address uncertainty quantification at all three of the aforementioned levels.

This talk is part of the Engineering Design Centre Seminars series.

Tell a friend about this talk:

This talk is included in these lists:

Note that ex-directory lists are not shown.

 

© 2006-2018 Talks.cam, University of Cambridge. Contact Us | Help and Documentation | Privacy and Publicity