Session: 02-06-01: Product and Sustainable Design

Paper Number: 117049

117049 - Optimization of Turbomachinery Design for S R - 30 Small Scale Gas Turbine Engine Using Machine Learning 

Gas turbine engines (GTE) have been used extensively in the aviation sector for several decades for propulsion. The Engine Development Lifecycle (EDL) consist of 5 stages: Thermodynamic Performance calculations to determine efficiencies of GTE through effective design of the components such as compressors, turbines etc.; Manufacturing processes and assembly of the components;  integration of the electrical, control and mechanical systems; Testing and fault diagnostics to ensure the design, structural and thermal performance; and Maintenance, emissions control, and engine health monitoring.  The interdependence of design and performance is the primary factor leading to extremely long design cycle times to obtain maximum efficiency of the GTE. To improve overall operational efficiency while keeping the design times nominal, the main goal of this paper is the development of a novel multi-physics optimization method to design turbomachinery components of the GTEs based on physics based thermodynamic performance computations using machine learning. Small scale gas turbine engines are known for their high power-to-weight ratio, high thermal efficiencies, and low emissions. However, the performance of small scale gas turbine engines can be affected by several factors, such as the compression ratio, turbine temperature, and the use of advanced materials and cooling systems. Combining experimental and numerical approach to increase the operational efficiency of the small-scale GTE SR-30 through probabilistic and physics based machine learning models will play a critical role in predicting the off-design and transient behavior of the engine with uncertainty quantification. In the experimental phase, the SR-30 is operated within the rotational speed of 40,000 to 80,0000 (RPM) using two fuels such as Jet-A and Kerosene. The experimental data obtained will assist in understanding the non-linear nature of the performance efficiencies based on component design of radial compressor and axial turbine. To mimic the experimental research, analytical models are developed based on the parametric geometric design blades in the compressor and turbine such as the inlet and outlet blade angles, chord, and the like. This high-fidelity numerical modeling coupled with experimental data for SR-30 GTE will enable prediction of off-design and transient performance parameters while optimizing the design to increase efficiency while minimizing losses. This small-scale turbine design optimization tool will be a holistic plug-and-play modular software to optimize the design and monitor performance for efficient, sustainable and cost-effective operation of SR-30 gas turbine engine. This tool will act as a preliminary design-performance tool to conduct a holistic evaluation of gas turbine engines.

Presenting Author: Sowmya Raghu UofSC

Presenting Author Biography: Ph.D. Candidate in Mechanical Engineering at the University of South Carolina

Authors:

Sowmya Raghu UofSC
Jamil Khan University of South Carolina