Civil aero-engine performance prediction using a low-order code and uncertainty quantification estimation

Abstract

In the last decades, the attention for pollutant emissions in the civil air transport field has grown up continuously. Especially considering the performances of current turbofan engines, even a modest increase in overall efficiency can lead to great benefits in terms of emissions reduction. Therefore, dedicated performance prediction tools are mandatory in order to carry out an estimation of such outputs. The aim of the present study is to develop a procedure devoted to a preliminary output prediction of an aero engine for civil transportation and an uncertainty quantification analysis based on main performance parameters. For the first step, following the strategy already adopted in previous work on this topic [1], the GEnX, a high-bypass turbofan engine, has been considered as the reference cases. The main design characteristics available from the constructor for this engine have been employed to model the engine with a 0-D numerical tool (ESMS), developed by the University of Florence [2]. Great efforts have been required in the modelling of the GE9X engine since few technical data are available currently. The estimated engine performances have been compared in the LTO cycle with the official data coming from ICAO [3], showing a good agreement. The second step, involving the UQ analysis, takes into consideration 4 specific input parameters that mainly affect the modelling and, thus, the performance outputs, of the engine. BPR, inlet massflow, fuel massflow and isentropic efficiency of the components are taken into consideration. A recently developed Uncertainty Quantification (UQ) methodology using DAKOTA have been employed in order to outline the results: thrust and T49 have been considered. Moreover, a sensitivity through Sobol's indices has been carried out in order to seek the most influencing parameters. More in detail, a polynomial chaos approach has been adopted. This numerical approach gives very accurate results, comparable to a Monte Carlo analysis, with a remarkable drop in term of computational effort required: only 16 evaluations are required in order to carry out a complete uncertainty analysis. Results will show how the efficiency dominates for the thrust whereas a more balanced situation is estimated for the T49.