About

The Center for Predictive Engineering and Computational Science (PECOS) is a research center in the Institute for Computational Engineering and Sciences (ICES) at the University of Texas at Austin. It is dedicated to the development and application of tools and techniques for making reliable computational predictions of complex systems. The challenge of making reliable predictions in such systems is that they typically involve multiple interacting physical phenomena acting at a wide variety of length and time scales. Computational models for such systems typically involve models of the the relevant phenomena with widely varying reliability and provenance, from empirical curve fits to well established fundamental theory. Furthermore, inputs to a system model, such as the environment in which it operates and various model and system parameters may be poorly or not known. Reliable predictions in such systems, therefore, requires determining whether the models and supporting data are sufficient for the prediction and its intended use.

To advance predictive computational modeling in complex systems, PECOS scientists, led by Prof. Robert Moser, pursue research in several broad areas as described briefly below.

Validation and Uncertainty Quantification: Reliable predictions  require that models and other assumptions be validated against relevant observations and that uncertainties be accounted for. PECOS is engaged in a broad program to advance techniques for validation and UQ, including: forward and inverse problems, formulation of uncertainty representations, especially for uncertainty due to   modeling errors, validation under uncertainty, and validation of the  reliability of predictions.

Applications to Complex Systems: Validation and UQ techniques as described above are being developed and applied in PECOS to a variety of systems of varying complexity. These include: atmospheric reentry vehicles, external aerodynamics, film cooling in gas turbines, combustion, magnetically confined plasmas, contaminant dispersion in porous media, batteries and super-capacitors, and oil and gas well drilling.

Turbulence Simulation and Modeling: Many complex systems involve fluid dynamic turbulence, which presents many computational, modeling and UQ challenges. Computational turbulence research at PECOS includes direct numerical simulation, and modeling for Reynolds averaged Navier-Stokes, large eddy simulation and hybrid approaches.  Also being pursued at PECOS is the formulation of representations for the uncertainty due to turbulence modeling errors.