Dipartimento di
Ingegneria Meccanica e Aerospaziale

Progetto PRIN 2022 n. 2022B2X937 -finanziato dall’Unione Europe Next Generation EU”
 
Titolo:NextGenSProDesT Next Generation Space Propulsion Design Techniques -
 
Abstract:
The device of interest is a LOX-CH4 Liquid Rocket Engine (LRE) equipped with a pintle injector, regarded as a natural choice when the mission profile requires a large throttling range. The project aims at leveraging advanced mathematical techniques to generate a reliable multi-fidelity model, built upon both experimental and numerical high-fidelity data and capable of embedding the key physical features of the latter. The novelty of the project lies in a mathematically sound workflow to exploit all the available experimental data and systematically compensating for the lack of experiments by means of high-fidelity numerical modeling. The resulting augmented-fidelity model will embed the uncertainty on the parameters and, once validated, will be exploited to generate a low-fidelity Response Surface Model (RSM) to be employed in Computer-Aided Engineering (CAE) Robust Design Optimization (DO). The results of the project have a strong translational impact on the entire aerospace propulsion and power generation industry.  The project’s focus is placed on three fundamental aspects to be accounted for in the numerical modeling of an LRE thrust chamber, namely, wall-turbulence interaction and heat flux estimation, spray breakup dynamics, chemical kinetics for fuel oxidation. As concerns the former issue, high-fidelity data are generated via wall-modeled LES calculations on both prototypical validation test cases and the full pintle-based thrust chamber configuration. The calibration of a RANS submodel is carried out through field inversion and machine learning paradigm. With regard to spray breakup dynamics, experimental data available in the literature are exploited in a Bayesian framework via uncertainty quantification (UQ) techniques to characterize uncertain breakup model parameters in terms of a posterior probability density function (PDF). As regards chemistry modeling aspects, high-fidelity data are provided by 0D reactor models and 1D flame configurations investigated through a detailed kinetic mechanism suitable for methane oxidation at high-pressure levels. Moreover, a computational singular perturbation (CSP) framework, featured with tangential stretching rate (TSR) analysis, is employed to generate a set of skeletal schemes. The latter is calibrated in a UQ-based Bayesian context, allowing for model error inference to account for the intrinsic discrepancy between the reduced scheme and the high-fidelity detailed mechanism. The entire calibration process results in the augmented-fidelity RANS solver, which is exploited for sensitivity analysis on the pintle injector configuration and to drive additional experimental campaigns to reduce the uncertainty of the models.
Lastly, starting from the augmented-fidelity model, the low-fidelity, cheap-to-evaluate RSM, is generated through artificial intelligence techniques and is subsequently employed in massive CAE design optimization campaigns.
 
Team di ricerca:
Pietro Paolo Ciottoli,
Jacopo Liberatori,
Leandro Lucchese
 
Partner:
Università degli Studi della Campania "Luigi Vanvitelli” (Prof. G. De Stefano) ,
Politecnico di TORINO (Prof. A. Ferrero)
   
Durata:
dal 28/09/2023  al 28/2/2026 24 mesi (prorogato)
 
Importo finanziato (DIMA):
75.840 €