Eventos Anais de eventos
COBEM 2023
27th International Congress of Mechanical Engineering
Combined Screening Sensitivity Analysis Applied to One-dimensional Model of a High-Performance S-CO2 Centrifugal Compressor
Submission Author:
Fernando Henrique Tiezzi Vergara , SP , Brazil
Co-Authors:
Fernando Henrique Tiezzi Vergara , Elóy Esteves Gasparin, Jurandir Itizo Yanagihara, Leandro Salviano
Presenter: Fernando Henrique Tiezzi Vergara
doi://10.26678/ABCM.COBEM2023.COB2023-0646
Abstract
A one-dimensional simulation is the first step on the development of centrifugal compressor devices. From that starting design point, it’s possible to reduce variables that are taken into account and have a good approximation for the most efficient geometry. Thus, saving time, money and achieving a better performance. Moreover, two mathematical methods were chosen to perform the screening Sensitivity Analysis (SA): Morris and Fractional Factorial (FF) methods. Both working with the same acting range of variables that, in this case, are eight (Blade thickness, number of blades, splitter blade length, shroud radius, axial height, impeller exit height, sizing parameter factor and specific diameter). After ranges are established, the method’s Design of Experiments (DoE) are computed in a python program (One-dimensional model) that returns as outputs, the parameters to be analyzed: isentropic efficiency, consumed power and pressure recovery. Then, all the data is stored and analyzed in order to identify which variable configuration returns the best efficiency. Furthermore, these two screening SA methods are combined, because the Morris method is uncapable to estimate the interactions between different factors due your nature of sampling, this way, the FF complements it in order to be compared with the SS-ANOVA method, which allows the evaluation of the screening method capacity of properly identifying interactions between variables, ensuring reliable and relatively cheap sample space representativity. Finally, the SA is able to find the most relevant variable for the system’s efficiency, which can demonstrate the future focus of further studies. Therefore, with these preliminary studies it is possible to have a reliable glance of the sample space behavior, and which variable is more relevant for the system. Showing the importance of a one-dimensional model simulation in models with high number of variables as a low-cost first step of centrifugal compressor’s design.
Keywords
One-dimesional Model, s-CO2 Centrifucal Compressors, Combined Screening Sensitivity Analysis.
