The difference between traditional experiments and CFD validation benchmark experiments

Computation Fluid Dynamics provides attractive features for design, and perhaps licensing, of nuclear power plants. The most important of these features is low cost compared to experiments. However, uncertainty of CFD calculations must accompany these calculations in order for the results to be useful for important decision making. In order to properly assess the uncertainty of a CFD calculation, it must be “validated” against experimental data. Unfortunately, traditional “discovery” experiments are normally ill-suited to provide all of the information necessary for the validation exercise. Traditionally, experiments are performed to discover new physics, determine model parameters, or to test designs. This article will describe a new type of experiment; one that is designed and carried out with the specific purpose of providing Computational Fluid Dynamics (CFD) validation benchmark data. We will demonstrate that the goals of traditional experiments and validation experiments are often in conflict, making use of traditional experimental results problematic and leading directly to larger predictive uncertainty of the CFD model.