Using Machine Learning Tools to Predict Compressor Surge and Stall

摘要

Phenomena such as surge and stall are some of the most critical problems facing the operation of compressors in traditional gas turbine fuel cell hybrid power systems. Due to the increase in volume that occurs during the operation of gas turbines, these dynamics are further magnified and proliferated throughout the compressor. This occurrence of stall and surge affects the performance, stability, reliability, and safety of the compressor, resulting in potential damage to the fuel cell. The hybrid performance (Hyper) facility at the U.S. Department of Energy's National Energy Technology Laboratory (NETL) was designed to evaluate the dynamic operation of gas turbine hybrid cycles. The facility simulates tests on a variety of hybrid gas turbines and fuel cell types with the goal of testing and developing these technologies before implementing them in a pilot scale fashion. Due to the presence of gas turbines and compressors in the facility, Hyper is subject to stall and surge. This study explores the utilization of machine learning tools to predict compressor stall and surge within Hyper.