A novel design-oriented computational framework based on wall-modeled large-eddy simulation (WMLES) coupled with a sharp-interface immersed boundary method (IBM) is proposed to enable rapid parametric mapping of unsteady aerothermal behavior in gas-turbine components. Rather than using scale-resolved simulations solely for detailed flow analysis, the present approach targets engineering outputs—compact, interpretable wake descriptors and surrogate response surfaces that can be queried within preliminary design and trade-study loops. The methodology is demonstrated on the LS89 transonic turbine cascade by varying two practically relevant drivers: inflow turbulence intensity and wall thermal condition. A GPU-optimized solver is used to generate a database of 64 WMLES cases at a per-case wall-clock turnaround that becomes compatible with systematic parametric campaigns—a regime previously accessible only to Reynolds-averaged-based workflows. The resulting scale-resolved fields are condensed into a hierarchy of spatial and temporal metrics representative of wake mean state and unsteady content, and these metrics are approximated through Chebyshev surrogate models defined over the parameter domain. The analysis indicates that wall thermal conditions predominantly control the mean thermodynamic wake state, whereas increasing inflow turbulence intensity redistributes unsteady pressure energy from high to intermediate Strouhal ranges and reduces spectral coherence. Overall, the proposed workflow provides a practical route from high-fidelity unsteady simulations to design-ready maps and response surfaces, bridging the gap between WMLES-level physics and parametric decision-making in gas-turbine design.
A design-oriented IBM–WMLES framework for rapid parametric mapping of turbine-cascade wake aerothermodynamics
De Vanna F.
;Benini E.
2026
Abstract
A novel design-oriented computational framework based on wall-modeled large-eddy simulation (WMLES) coupled with a sharp-interface immersed boundary method (IBM) is proposed to enable rapid parametric mapping of unsteady aerothermal behavior in gas-turbine components. Rather than using scale-resolved simulations solely for detailed flow analysis, the present approach targets engineering outputs—compact, interpretable wake descriptors and surrogate response surfaces that can be queried within preliminary design and trade-study loops. The methodology is demonstrated on the LS89 transonic turbine cascade by varying two practically relevant drivers: inflow turbulence intensity and wall thermal condition. A GPU-optimized solver is used to generate a database of 64 WMLES cases at a per-case wall-clock turnaround that becomes compatible with systematic parametric campaigns—a regime previously accessible only to Reynolds-averaged-based workflows. The resulting scale-resolved fields are condensed into a hierarchy of spatial and temporal metrics representative of wake mean state and unsteady content, and these metrics are approximated through Chebyshev surrogate models defined over the parameter domain. The analysis indicates that wall thermal conditions predominantly control the mean thermodynamic wake state, whereas increasing inflow turbulence intensity redistributes unsteady pressure energy from high to intermediate Strouhal ranges and reduces spectral coherence. Overall, the proposed workflow provides a practical route from high-fidelity unsteady simulations to design-ready maps and response surfaces, bridging the gap between WMLES-level physics and parametric decision-making in gas-turbine design.
Pubblicazioni consigliate
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
