Digital Twin Framework application for Aeronautical Hydrogen Microinjectors Design
Please login to view abstract download link
The rise of alternative fuels leads to numerous new injection technologies, for instance micro-injectors [1], which rely on the creation of multiple miniaturized flamelets in order to reduce NOx production. From a design and engineering perspective, new sets of tools of varying fidelity are needed to make the design-screening step faster and more exhaustive. A reduced-order model (ROM) based on the OpenMeasure [2] library [https://github.com/albertoprocacci/OpenMEASURE] has been implemented coupled with a Design Of Experiments (DOE) library, based on Lagun [3], in order to create a digital twin of hydrogen micro-injectors. The ROM is based on the use of Proper Orthogonal Decomposition (POD) for data compression, manifold alignment for the transfer of fidelity and Co-Kriging for interpolation. This approach has been carried out on 24 designs, where several geometrical parameters are varied, e.g. number of fuel injection holes, aspect ratio, etc. The prediction of NOx via the reduced model was assessed using 44 RANS simulations, hence establishing a database of micro-injector behavior. Similarly, several operating conditions have been varied, i.e. pressure (atmospheric and high pressure), equivalence ratio and fuel mass flow rate. A selection of model inputs was made based on an assessment of the model's predictive accuracy using the Co-Kriging estimation method. Experimental measurements of pollutant emissions at atmospheric pressure were conducted including emissions at the chamber exit and OH-PLIF (Planar Laser Induced Florescence) to visualize the flame. This data was then used to correct the predicted behavior and enrich the low-order model.
