Direct numerical simulation of wall-bounded turbulent flows with adverse and favorable pressure gradients

• Name: Direct numerical simulation of wall-bounded turbulent flows with adverse and favorable pressure gradients , TBLDNS
• EuroHPC machine used: Leonardo
• Topic: Engineering and Technology – Mechanical Engineering

Overview of the project

This research project investigates the structure and dynamics of wall-bounded turbulence in adverse and favorable pressure gradient (APG and FPG) turbulent boundary layers (TBLs) using direct numerical simulation (DNS). An APG is a pressure gradient that decelerates the flow, potentially leading to flow separation from the aerodynamic surface, whereas an FPG accelerates the flow, increasing skin friction drag. The overall research goal is to develop and improve knowledge and theory that help describe and model APG/FPG TBLs. The long-term objectives of this research are to revolutionize the design of energy generation and transport platforms that operate in APG and FPG environments and to develop active flow control systems that increase the operational envelope. This will improve energy efficiency over a wide range of operating conditions for the flow over aircraft wings, wind turbine blades, and any form of turbomachinery. Improvements in the performance of such systems will result in more efficient and cleaner power generation, reduced fuel usage, and lower CO2 emissions. These objectives can only be fulfilled with the help of very detailed flow data obtained numerically with DNS using HPC.

How did EPICURE support the project and what were the benefits of the support?

“The primary challenge was optimizing our in-house Fortran solver for a specific high-performance computing architecture. We lacked comprehensive knowledge of machine-specific optimization procedures, appropriate compiler flags, and optimal library configurations for the Leonardo DCGP environment.

The support from the EPICURE team was highly beneficial for our project. They transitioned our code’s communication library from OpenMPI to Intel MPI and switched all math and FFT operations to Intel MKL. These changes, combined with compiler optimizations, resulted in a roughly 15% total performance gain in terms of speed and allowed us to maximize our utilization of the Leonardo DCGP system. While we cannot provide a specific estimate for energy consumption, this improvement in efficiency significantly speeds up our production runs and will remain a permanent part of our workflow for all future simulations.” – Mehmet Ali Yesildag

Additional references

Simens, M. P., Jiménez, J., Hoyas, S., & Mizuno, Y. (2009). A high-resolution code for turbulent boundary layers. J. Comp Physics, 228(11), 4218-4231.

Borrell, G., Sillero, J. A., & Jiménez, J. (2013). A code for direct numerical simulation of turbulent boundary layers at high Reynolds numbers in BG/P supercomputers. Computers&Fluids, 80, 37-43.

Gungor, T. R., Gungor, A. G., & Maciel, Y. (2024). Turbulent boundary layer response to uniform changes of the pressure force contribution. Journal of Fluid Mechanics, 997, A75. https://doi.org/10.1017/jfm.2024.579

Gungor, T. R., Gungor, A. G., & Maciel, Y. (2025). The three effects of the pressure force on turbulent boundary layers. International Journal of Heat and Fluid Flow, 116, 109898. https://doi.org/10.1016/j.ijheatfluidflow.2025.109898

Yesildag, M. A., Gungor, T. R., Gungor, A. G., & Maciel, Y. (2025). Spatial features of Reynolds-stress carrying structures in turbulent boundary layers with pressure gradient. International Journal of Heat and Fluid Flow, 116, 109883. https://doi.org/10.1016/j.ijheatfluidflow.2025.109883