Initial Development of Statistically Based Validation Process for Computational Simulation (original) (raw)

50th AIAA Aerospace Sciences Meeting including the New Horizons Forum and Aerospace Exposition, 2012

Abstract

The initial development of a statistically-based process for validation of computational experiments is presented. The focus of this newer methodology is Uncertainty Quantification (UQ), Sensitivity Analysis (SA), and variance reduction. The two types of uncertainty (aleatory and epistemic) are incorporated into the methodology. For this investigation, the behavior of the simulation inputs is presumed unknown (aleatory due to lack-of-knowledge) and is therefore modeled using equally-probable uniform distributions. These distributions have known and quantifiable uncertainties that are used to determine the sampling residuals that are expected to follow a random normal distribution (epistemic with known mean and variance). Statisticallybased sample sizes and uncertainty propagation limits are determined based upon a priori tolerance specifications as well as Type I and Type II risks. Newer quasi-random sampling procedures are demonstrated to be superior to classical pseudo-random sampling procedures. A simple example is presented that outlines the methodology that is extensible to all validation processes.

john doty hasn't uploaded this paper.

Let john know you want this paper to be uploaded.

Ask for this paper to be uploaded.