Scale-resolving Simulations ​for Innovations in Turbomachinery Design

Informacje na temat projektu

Sci-Fi-Turbo

Identyfikator umowy o grant: 101138080

DOI

10.3030/101138080

Data podpisania przez KE 24 Listopada 2023

Data rozpoczęcia 1 Stycznia 2024

Data zakończenia 30 Czerwca 2027

Finansowanie w ramach

• Climate, Energy and Mobility

Koszt całkowity

€ 3 436 753,75

Wkład UE

€ 3 436 753,75

Koordynowany przez

DEUTSCHES ZENTRUM FUR LUFT - UND RAUMFAHRT EV
Germany

Periodic Reporting for period 1 - Sci-Fi-Turbo (Scale-resolving Simulations ​for Innovations in Turbomachinery Design)

Okres sprawozdawczy: 2024-01-01 do 2025-06-30

Podsumowanie kontekstu i ogólnych celów projektu

The transition to greener aviation is a global priority, but traditional industrial design methods for turbomachinery—the core of jet engines—are a major bottleneck. These methods rely on simplified computational models called Reynolds-Averaged Navier-Stokes (RANS) models, which are good for general design but lack the precision needed for next-generation, ultra-fuel-efficient engines. The SciFiTurbo project was launched to overcome this barrier by creating a new, hybrid design approach.
Our core objective is to integrate highly accurate, but computationally intensive, Scale-Resolving Simulations (SRS) into standard industrial design workflows. We then combine this with machine learning (ML) to create a powerful, new design pipeline. This hybrid approach uses a small number of targeted SRS runs to "teach" data-driven models, which can then spread this high-fidelity insight across many faster, lower-cost simulations. By doing this, we aim to drastically shorten development cycles, reduce the need for expensive physical testing, and cut costs. This will directly enable the development of new propulsion technologies needed to meet the EU's climate-neutral aviation goals for 2050.

Prace wykonane od początku projektu do końca okresu sprawozdawczego oraz najważniejsze dotychczasowe rezultaty

During the first reporting period, the SciFiTurbo consortium successfully implemented and demonstrated a fully automated, end-to-end design workflow using Scale-Resolving Simulations (SRS) for a representative compressor cascade test case. This work combined technical advancements in high-order SRS solvers, such as adaptive time stepping, with workflow automation, including mesh adaptation, real-time visualization of the simulation, and automated convergence monitoring.
A major achievement was the establishment of a multi-fidelity optimization chain that uses the new hybrid approach, which we are soon to test on the compressor cascade case. We've confirmed on existing experimental data that SRS provides a significant improvement in aerodynamic accuracy over traditional Reynolds-Averaged Navier–Stokes (RANS) models. Crucially, we demonstrated that a small number of targeted SRS runs can effectively guide the optimization process toward superior designs. Our demonstrations of combining SRS, RANS, and machine learning (ML) indicate this hybrid approach to be a highly effective compromise between accuracy and computational cost—a trade-off previously unattainable in industrial settings. We are currently preparing to validate these methods on full-scale 3D fan-stage components and with experimental data.

Innowacyjność oraz oczekiwany potencjalny wpływ (w tym dotychczasowe znaczenie społeczno-gospodarcze i szersze implikacje społeczne projektu)

SciFiTurbo is making significant progress in moving high-fidelity simulation from a niche academic tool to a practical industrial solution by establishing a new, more efficient paradigm for aero-engine design.
• Operational High-Fidelity Simulation: We've successfully integrated high-fidelity Scale-Resolving Simulations (SRS) into an automated, end-to-end design chain for compressor cascades. This new workflow has demonstrated its ability to meet industrial time constraints for setting up, executing, and analysing complex simulations. A key next step will be to extend these capabilities to full 3D applications, which we anticipate will make SRS a usable decision-making tool during normal design and optimization cycles, rather than just an isolated proof-of-concept.
• Effective Small-Sample Strategies: Our work to date has shown that a very small number of carefully chosen high-fidelity simulations can be used to successfully correct conventional Reynolds-Averaged Navier–Stokes (RANS) optimizations. This progress has produced far better results with an acceptable increase in computational cost, thus proving the potential of a new, highly efficient design strategy.
• Practical Hybrid Workflows: By combining targeted SRS with fast surrogate predictors and machine learning-based turbulence corrections, we are working to achieve a favourable trade-off between predictive accuracy and computational expense that was previously impossible. Our initial demonstrations indicate this approach will allow for more ambitious designs with a higher probability of success.
To translate these advances into routine industrial practice, the consortium has identified key enablers that will be addressed in the next phase of the project: standardized workflow drivers and open benchmark datasets. Addressing these needs will accelerate industrial uptake and unlock the environmental and economic benefits of this new approach.