NNSA PRISM Center Workshop: "Sensitivity Analysis, Uncertainty Propagation and Validation...."
January 14 @ 8:00 AM - - January 15 @ 5:00 AM - MRGN 129
Ben Blackwell, Blackwell Consulting, Albuquerque, NM
Kevin Dowding, Sandia National Laboratories, Albuquerque, NM
Thursday & Friday, January 14-15, 2010 / 8:30 a.m. - 4:30 p.m.
Burton D. Morgan Center for Entrepreneurship / Room 129
Engineers are faced with the task of producing robust designs more cheaply in order to face the challenges of economic competitiveness. This means that the parameter space that controls the design must be fully explored. Single point calculations are no longer considered the norm. Sensitivity analysis is the formal technique of determining those parameters in a system (computational model) that controls its performance. It identifies those parameters that are important as well as those that are unimportant. The course will specifically focus on the following techniques for determining sensitivity information: differentiation of analytical models, finite difference of computational models, complex step method, software differentiation, sensitivity equation methods, adjoint methods, and sampling methods (Monte Carlo and Latin Hypercube). Practical sensitivity analysis and uncertainty propagation examples will be taken from our experiences with these methods. Advantages and disadvantages of each method will be presented. Techniques for propagating uncertainty through computational models and how this computational uncertainty interacts with experimental uncertainty and model validation will be presented.
The propagation of uncertainty naturally feeds into the validation of computational models. From ASME V&V 20-2009, validation is “the process of determining the degree to which a model is an accurate representation of the real world from the perspective of the intended uses of the model.” Validation requires a comparison between experimental data (along with its uncertainty) and a computational simulation of the experiment (along with its uncertainty). An overview of validation and description of the approach in V&V-20 are discussed. The validation of a hypothetical heat transfer experiment of a fin-tube heat exchanger is presented. Additional, a practical example demonstrating validation is discussed.
Dr. Blackwell is retired from Sandia National Laboratory where he was a Distinguished Member of the technical staff. He spent over 36 years at Sandia in the areas of computational and experimental heat transfer. Since his retirement, he has consulted with Sandia on the development of ablation codes and verification of computational heat transfer software and in the aerospace community on inverse methods for estimating heat flux from temperature measurements. He has published extensively and is a fellow of ASME and associate fellow of AIAA.
Dr. Dowding is a Principal Member of the technical staff at Sandia National Laboratories. He has worked the past 12 years on sensitivity analysis, uncertainty analysis, and application of these analyses to the validation of computational models. He has participated and lead projects to apply these analyses to complex thermal-fluid modeling applications and as a part of model-based risk assessment. He is active in ASME and AIAA.
Both Drs. Blackwell and Dowding are co-authors of ASME V&V 20-2009, “Standard for Verification and Validation in Computational Fluid Dynamics and Heat Transfer.
- Mari-Ellyn Brock