After receiving its latest SBIR award from NASA, SoftInWay will begin a new R&D project to use artificial intelligence to improve the methodologies and workflows currently used to develop highly complex turbomachinery.
Compressor efficiency is one of the most important factors in developing an efficient gas turbine engine. Often, it can take years of designing, building, and testing a compressor before a proper design is found which operates both reliably and efficiently. With the AI and machine learning technology being developed by SoftInWay, optimizing a compressor, or any turbomachine for that matter, to function at optimal performance will be possible in weeks instead of years.
Creating an autonomous program that can accurately optimize a compressor’s operating efficiency and generate performance maps is a daunting task. At the heart of the program will be the AxSTREAM platform, more specifically, the AI capabilities being created in AxSTREAM ION. This new AI-based workflow will be trained using AxSTREAM, but it can also be trained using proprietary software and third-party codes such as those at NASA. These programs will be able to automate performance data generation for compressors in this test case. When subsequently combined with data gathered from test rigs, the autonomous AI program will be able to generate accurate turbomachinery performance data, with minimal error percentages; significantly shortening the time required to design a complex machine such as a multistage compressor.
This technology will go beyond axial compressor R&D, into applications such as turbopump design for liquid rocket engines, as well as turbocharger design. While all of these machines are very different, the engineers creating them all share a common challenge, which is being able to create more powerful, efficient, designs which operate optimally in ever-shortening time frames.