Axial Compressor Map Generation Leveraging Autonomous Self-Training AI

NASA is looking for improvement in aeropropulsive power density and efficiency in support of its Strategic Thrust in the area of Ultra-Efficient Subsonic Transports, focusing on small core turbofan engines for next-generation and future large commercial transport aircraft. The trend in the design of modern gas turbine engines is for ever-increasing cycle efficiency and reduced specific fuel consumption. To achieve these engine cycle efficiency goals, the low and high-pressure compressors (HPC) are pushed to ever-increasing levels of pressure ratio. Increasing levels of compressor pressure ratio results in higher rotor tip relative Mach number in the HPC front stages, and consequently steeper performance characteristic maps. The compressors with steep characteristics typically require variable geometry inlet guide vanes as well as variable stators in the first few stages to provides the desired performance and stability in an engine system. The design and development time of a modern high-pressure compressor with variable geometry can take years of design-build-test iterations which includes testing a large number of possible reset angles of the variable vanes. Determining the optimal combination of vane angle resets that will provide the desired compressor performance in an engine system environment is a time consuming and expensive part in the development of high-pressure compressors. It is proposed to address the optimization of the variable geometry reset angle schedules with the use of the innovative autonomous AI technology. The AI-based performance prediction model can be easily incorporated inside of the system analysis tool and reliably predict the performance with high accuracy across the entire operating range of compressor even with multiple variable guide vanes and thus helping to approach true optimal engine performance and reduce the chances of additional expensive design iterations in real-life projects.