Probabilistic Methodology for Establishing Test Requirements in a Gas Turbine System Development and Demonstration Program

There is a pressing need in the gas turbine community for advanced simulation-based methodologies that can reduce test requirements for gas turbine component/engine qualification. Currently, no such simulation methodology exists that is statistically accurate and employs high-fidelity simulations to reduce hardware testing. In this SBIR, CFDRC will develop such a simulation-based methodology. In Phase I, a probabilistic approach using basic statistical techniques (response surfaces and Monte Carlo simulations) was developed to show the potential of test reduction through simulation. Feasibility of the approach was demonstrated using conceived augmentor simulation and experimental data. The need for highly accurate simulations was shown, and will be addressed in Phase II. In Phase II, cutting-edge CFD technologies of 1) Second Order Second Moment (SOSM) method, and 2) non-linear, sensitivity derivatives using Automatic Differentiation (AD), will be developed to propagate uncertainties through the simulation. These technologies will be implemented into the highly accurate DREAM code being developed under AFRL sponsorship, thus creating the much-needed infrastructure in CFD to meet the accuracy requirements of probabilistic methodology. For initial demonstration on a simple engine sub-component, the AFRL Augmentor High-Impact-Technology (HIT) rig will be analyzed. Statistical analysis of simulation and limited test data will be performed to determine hardware test reduction for the same operational and reliability confidence level. As a final demonstration on the engine component level, CFDRC will repeat the probabilistic process on a Williams International (WI) augmentor.