Global fuel consumption is continuing to climb, with commercial airlines alone reaching an all-time high of 363 billion litres in 2019. Increasing the efficiency of an aircraft by even 0.1 per cent would have a massive global impact on fuel demand, costs, and the environment.
Richard Sandberg and a team at the University of Melbourne are working towards this impact, using high-fidelity simulation approaches to examine the effects turbulence has on aircraft engine efficiency, ultimately creating cleaner methods of air travel.
Making aircrafts more efficient is a large-scale problem that demands an equally large-scale, significant solution. Richard Sandberg and his team at the University of Melbourne are taking on the challenge, working to better understand turbulent flows in jet engines and how they affect efficiency and noise generation.
Sandberg’s work combines numerical methods and high-performance computing to predict gas flows in turbine engines and translate the data generated into predictive models.
This work is unique in that it is conducted and resolved on full-scale gas turbine flows. The type of high fidelity work where turbulent flow is represented by simulation requires tremendous computing power. Performing calculations at real engine conditions, rather than at half-size model scale, drives the computational cost up eight times.
The team’s predictive modelling methods use around 100 million computer hours per year — the equivalent of approximately 3,000 years on conventional desktop computers.
Backed by Magnus’ computing power, Sandberg’s high-performance computing(HPC)-optimised compressible fluid solver is able to produce predictions of all the detailed turbulent motions that affect efficiency. The data generated by these computational fluid dynamics simulations is vast, with a single simulation generating tens of terabytes, which need to be downsampled in order to store and subsequently deliver research findings.
“I’m focused on solving equations that are very non-linear and complex that we have not been able to find general solutions for to date. With this technology, we can now use high-fidelity simulation approaches to conduct world-leading research for industrially-relevant problems,” explains Sandberg.
Evolving such complex, large-scale machines requires a focus on the most minute details. As a leader in the study of turbulent flows, Sandberg’s team is paving the way for further understanding and developing the way gas turbine engines perform under various conditions.
The data Sandberg generates is used to obtain direct physical insight for his own project, as well as help improve industry models. Sandberg’s team has developed machine learning tools that translate the large-scale data he generates into plug-and-play models. In fact, these tools have already attracted the attention of the US Navy, General Electrical, and Mitsubishi Heavy Industries.
With a secret superpower in supercomputers, Sandberg has been able to contribute significant outcomes to the complex issue of turbulent flows. It’s just the start of a new form of research in the field, but the early results from his simulations are paving the way for huge global impact.