Abstract

Many current methods of analysis and design are performed using ad-hoc heuristic and/or parametric methods are inefficient, but more importantly—may be ineffective as well. Every experiment is designed—unfortunately most are improperly designed based upon lack of knowledge of the physics and/or lack of statistical knowledge. Several examples will be introduced that demonstrate the benefit of using physics-based design (1st & 2nd Laws of Thermodynamics) coupled with advanced statistical techniques (Design of Experiments (DoX), inferential statistics, response surfaces, canonical and/or orthogonal designs, hyperspace visualization). An overview of Design of Experiments (DoX) will be provided with emphasis on data-driven statistical models from experiments as well as simulation-based statistical models relevant to AEDC. A mechanical system (steam plant), an electrical system (simple circuit), and an aerospace system (supersonic turbojet and wing section) will be used to demonstrate the statistically-based analysis, design, and optimization process.

Throughout the process, emphasis is placed on correctly employing physics (2nd Law in particular) as well as statistically-designed analysis points for increased leverage of information after the data is taken. The focus in the statistical modeling is to gain knowledge, not gather data. Combing the proper physics with the statistical design and analysis of the information provides a unique and powerful interpretation of the data, providing maximum utility of the information with minimum number of data points required. This information is further leveraged with advanced visualization techniques to enable the entire design and/or analysis space to be simultaneously interpreted.

The processes overviewed are easily adapted to existing systems, such as those at AEDC. Employing advanced statistical techniques with physics-based modeling and simulation can provide insights not possible with current analysis techniques. The goal of applying these techniques is to improve the efficiencies of large power systems, saving energy (money), time, as well as reducing the carbon footprint of these systems.

Please see the Seminar Announcement for further details.