Extrapolation-Based Discretization Error and Uncertainty Estimators
Tyrone S. Phillips and Christopher J. Roy
Several extrapolation-based error and uncertainty estimators were evaluated using various Euler solutions. Metrics used include the effectivity index (accuracy of the discretization error estimate relative to the true error) and conservativeness (fraction of cases where the uncertainty estimate encompasses the true error).
The results led to the development of a new uncertainty estimator which was evaluated using a new set of 3D, unstructured, Euler and Navier-Stokes solutions.
Key Findings
- A global order of accuracy should be used instead of locally calculated values to reduce scatter and improve conservativeness
- The proposed GCI-Δp uncertainty estimator meets the goal of 95% conservativeness for all evaluated data
References
- T. S. Phillips and C. J. Roy, “Evaluation of Extrapolation-Based Discretization Error and Uncertainty Estimators,” AIAA Paper 2011-215, 49th AIAA Aerospace Sciences, Jan. 4-7, 2011, Orlando, Florida.
- T. S. Phillips and C. J. Roy, “A New Extrapolation-Based Method for Estimating Grid-Related Uncertainties in CFD,” American Society of Mechanical Engineers (ASME) Verification and Validation Symposium, May 2-4, Las Vegas, NV.
- T. S. Phillips and C. J. Roy, “Evaluation of Extrapolation-Based Discretization Error and Uncertainty Estimators,” manuscript submitted to the Journal of Fluids Engineering, April 2012.
- A New Extrapolation-Based Uncertainty Estimator for Computational Fluid Dynamics," extended abstract submitted to the 51st AIAA Aerospace Sciences Meeting, 7 - 10 January 2013, Grapevine, TX.