High-fidelity simulations of turbomachinery applications

The long-term objective of this high-fidelity simulation program is to generate the fundamental understanding and low-order design tools required to develop the next generation of efficient and environmentally friendly turbines for power generation and air transport. Overall, our research team is focused on shedding light on the most significant loss generation mechanisms inside of gas turbines, how to further improve turbine cooling and identifying and addressing noise generation mechanisms of jets and wind turbine blades. In 2019, high-fidelity simulations have been conducted of fluid flows occurring in turbomachinery applications with an in-house code specifically designed and optimised to exploit the performance of supercomputers. We have conducted simulations at realistic engine conditions, giving us an unprecedented understanding of the physical phenomena occurring in these flows. The data is used to help assess the current low-order models that are employed in an industrial environment. The data also serves as a gold-standard database for developing new and improved models based on machine-learning approaches.

Principal investigator

Richard Sandberg richard.sandberg@unimelb.edu.au
Magnifying glass

Area of science

Engineering, Geosciences, Information And Computing Sciences

Systems used

Magnus, Zeus and Topaz

Applications used

Partner Institution: University of Melbourne| Project Code: bq2

The Challenge

Australia relies heavily on gas turbines (GTs) for propulsion- and power-generation. Given the very large installed base nationally, and worldwide, any GT efficiency increase has significant potential to reduce fuel burn and environmental impact. One of the keys to improving the operating efficiency of GTs is the ability to understand and predict the detailed flow behaviour inside of the engine. The challenge is that the flow inside an engine is mostly turbulent, and turbulence is a highly chaotic process that is extremely hard to predict. Today’s design of gas turbines is driven largely by computational fluid dynamic (CFD) modelling. However, some of these turbulence models are really only accurate in “known” situations, where the flow behaviour is well recognised and the models have already been calibrated. If off-design conditions or more radical novel concepts are to be considered, these models cannot be trusted. In addition the accurate prediction of the turbulent heat transfer is a key design consideration for ensuring survival of key turbine components and hence reliability of turbines. This research also aims to improve heat transfer models that are even less reliable and mature than current turbulence models.

The Solution

To avoid model uncertainty, the governing equations of fluid flow can be solved directly in so-called direct numerical simulations that resolve all scales of turbulence without assumptions. To resolve all the scales of turbulence, from large to small eddies, requires extremely large and costly simulations, with the number of unknowns exceeding 1,000 trillion values. However, the reward for performing these very large simulations is that a detailed and reliable representation of the flow and heat transfer inside an engine can be obtained. The data can then be used to gain new insight into fundamental physical mechanisms and to assess and develop new models. Prof Sandberg’s group at the University of Melbourne has developed a cutting-edge flow solver optimized for efficient use of modern high-performance computers that is one of the prerequisites for conducting direct numerical simulations of gas turbine flows.

The Outcome

Model-free direct numerical simulations of turbulent flow and heat transfer with relevance to gas turbines are only possible on very large supercomputers. The access to Magnus has allowed us to conduct a range of high-fidelity simulations that would not have been possible on local resources. Various cases have been studied, from looking at loss mechanisms in high-pressure turbines, in particular, the effect of inflow perturbations and/or Mach number variations on loss generation, impinging jets for turbine cooling, turbulent flow and heat transfer properties in cooling slot flows as function of slot thickness and blowing ratio, to the sensitivity of flow speed on noise generation of general airfoils.

Simulation of flow over an aerofoil: Background contours in black and white show acoustic waves emanating from the aerofoil. Coloured iso-contours visualise flow features producing the noise.

List of Publications from this project

Zhao, Y., Akolekar, H.D., Weatheritt, J., Michelassi, V., Sandberg, R.D. 2020, “Turbulence Model Development using CFD-Driven Machine Learning” accepted for publication in Journal of Computational Physics on 19/03/2020.
Deuse, M., Sandberg, R.D., 2020, “Different noise generation mechanisms of a controlled diffusion aerofoil and their dependence on Mach number” accepted for publication in Journal of Sound and Vibration on 09/03/2020.
Lav, C., Philip, J., Sandberg, R.D. 2020, “Compressible plane turbulent wakes under pressure gradients evolving in a constant area section” accepted for publication in Journal of Fluid Mechanics on 01/03/2020.
Zhao, Y., Sandberg R.D., 2020 “Using a new Entropy Loss Analysis to Assess the Accuracy of RANS Predictions of an HPT Vane”, accepted for publication in Journal of Turbomachinery, 10/02/2020.
Deuse, M., Sandberg, R.D., 2020, “Implementation of a stable high-order overset grid method for high-fidelity simulations” accepted for publication in Computers & Fluids on 18/01/2020.
Zhao, Y., Sandberg, R.D. 2020, “Bypass transition in boundary layers subject to strong pressure gradient and curvature effects” Journal of Fluid Mechanics, vol. 888. A4-1, doi:10.1017/jfm.2020.39
Otero, J.J., Sandberg, R.D. 2020, “Compressibility effects on heat transfer in turbulent impinging jets”, Journal of Fluid Mechanics, vol. 887, A15. doi:10.1017/jfm.2020.5.
Ma, M.-C., Talei, M., Sandberg, R.D., 2020, “Direct Numerical Simulation of turbulent premixed jet flames: Influence of inflow boundary conditions” accepted for publication in Combustion and Flame on 13/11/2019.
Weatheritt, J., Zhao, Y., Sandberg, R.D., Mizukami, S., Tanimoto, K. 2020, “Data-driven scalar-flux model development with application to jet in cross flow”, International Journal of Heat and Mass Transfer 147, 118931
Sandberg, R.D. and Wheeler, A.P.S., 2019 “Effect of Trailing-Edge Boundary Conditions on Acoustic Feedback Loops in High-Pressure Turbines”, Journal of Sound and Vibration 461, p.114917.
Akolekar, H., Sandberg, R.D., Hutchins, N., Michelassi, V., Laskowski, G., 2019, “Machine-Learnt Turbulence Closures for LPTs with Unsteady Inflow Conditions” Journal of Turbomachinery 141(10).
Weatheritt, J., Sandberg, R.D., 2019 “Improved Junction Body Flow Modeling Through Data-Driven Symbolic Regression”, Journal of Ship Research, 63(4), pp. 283-294, https://doi.org/10.5957/JOSR.09180053.
Sandberg, R.D., Michelassi, V. 2019, “The current state of high-fidelity simulations for main gas path turbomachinery components and their industrial impact”, Flow, Turbulence and Combustion, pp.1-52.
Lav, C., Philip, J., Sandberg, R.D. 2019, “A Framework to Develop Data-Driven Turbulence Models for Flows with Organised Unsteadiness”, Journal of Computational Physics, 383, pp.148-165.
Akolekar, H.D., Weatheritt, J., Hutchins, N., Sandberg R.D., Laskowski, G., Michelassi, V., 2019 “Development and Use of Machine-Learnt Algebraic Reynolds Stress Models for Enhanced Prediction of Wake Mixing in LPTs”, Journal of Turbomachinery, 141(4), p.041010-1.

Peer-reviewed proceedings:
Zhao, Y., Sandberg R.D., 2020 “High-fidelity Simulations of a High-pressure Turbine Vane: Effect of Exit Mach Number on Losses”, ASME IGTI, GT2020-14445.
Akolekar, H., Zhao, Y., Sandberg R.D., Pacciani, R., 2020 “Integration of Machine-Learning and Computational Fluid Dynamics to Develop Turbulence Models for Improved Turbine-Wake Mixing Prediction”, ASME IGTI, GT2020-14732.
Talei, M., Ma, D.M.C., Sandberg R.D., 2020 “Data-Driven Combustion Modelling for a Turbulent Flame Simulated with a Computationally Efficient Solver”, ASME IGTI, GT2020-14843.
Lav, C., Sandberg R.D., 2020 “Unsteady simulations of a trailing-edge slot using machine-learnt turbulence stress and heat flux closures.”, ASME IGTI, GT2020-14398.
Saini, D., Sandberg R.D., 2020 “Large Eddy Simulations of High Rossby Number Flow in the High-Pressure Compressor Inter-Disk Cavity”, ASME IGTI, GT2020-14463.
Otero, J.J., Sandberg R.D., Tanimoto, K., Mizukami, S., 2020 “High-Fidelity Simulations of Multi-Jet Impingement Cooling Flows”, ASME IGTI, GT2020-14728.
Akolekar, H., Zhao, Y., Sandberg, R.D., Hutchins, N., Michelassi, V. 2019, “Turbulence Model Development for Low & High Pressure Turbines Using a Machine Learning Approach”, International Symposium for Air-Breathing Engines, ISABE-2019-24010, Canberra, Australia, Sept 23-27, 2019.
Zhao, Y., Sandberg R.D., 2019 “Using a new Entropy Loss Analysis to Assess the Accuracy of RANS Predictions of an HPT Vane”, ASME IGTI, GT2019-90126.
Lav, C., Philip, J., Sandberg R.D., 2019 “A New Data-Driven Turbulence Model Framework for Unsteady Flows Applied to Wall-Jet and Wall-Wake Flows”, ASME IGTI, GT2019-90179.

Complex flow physics in High-pressure turbine simulations with turbulence generating bars at the inlet
Snapshot of temperature field from simulations of cooling flows with different slot thickness and blowing ratios