Predictive Low-Cost Large Eddy Simulation Capability for Fan and Open Rotor Noise

The research proposed addresses substantial capability advancements within the newly developed immersed boundary Wall-Modelled Large Eddy Simulation (WMLES) method for turbomachinery noise predictions. This will enable accurate and computationally efficient broadband aeroacoustic predictions for fan and open rotors. Substantial advancement of the current state-of-the-art is anticipated; the Phase-I proposal includes a WMLES demonstration of "overnight" R4 Rotor LES with farfield noise predictions using very modest computational resources - 2x NVIDIA RTX 4090 GPUs on an engineering workstation with aggressive performance and aero-acoustics accuracy targets. Additionally, the typically cumbersome pre-processing workflow starting with a CFD-ready CAD to the start of the LES will be entirely automated. This includes automated (but user-guided) surface and volume grid generation. In-situ, post-processing for flow visualization as well as in-situ (streaming) far-field acoustic propagation via the FWH analogy will be included as part of the demonstration. Two major advancements are needed in order to accomplish these goals: a) computationally efficient sliding mesh capability for overset rotating Cartesian grids, and b) time-step sub-cycling to mitigate the small time-steps required to resolve the flow in the rotor tip clearance. Both algorithmic enhancements are critical in order to maintain the advantages of the Cartesian immersed boundary WMLES capability for turbomachinery applications These advancements would expand the targeted market for Volcano ScaLES and make it available to US OEMs such as GE Aerospace, Pratt & Whitney, Honeywell, and RTX Corporation.