The number of air travelers and flights around the world is soaring. Peter Vincent wants to keep the resulting aircraft emissions from taking off.
Vincent and his team at Imperial College London are making it easier for manufacturers to design and test more environmentally efficient aircraft.
“The issue has been recognized on both sides of the Atlantic, and new technology is urgently required to meet the challenge,” said Vincent, who is a Reader and an Engineering and Physical Sciences Research Council Fellow in the college’s Aeronautics Department.
Lighter planes emit lower amounts of greenhouse gases, and many designers have zeroed in on reducing the weight of jet engine turbines. The researchers created open source software that lets designers use GPU-powered supercomputers to more efficiently and accurately simulate how new designs perform.
Emissions Could Triple
This work earned Vincent and his team a spot as one of six finalists for the ACM Gordon Bell Prize, considered the “Nobel Prize of supercomputing.” Awarded annually at the Supercomputing conference, the prize recognizes outstanding scientific achievement on the biggest and most powerful supercomputers.
Vincent says the aviation industry must develop greener aircraft as it grows. “It’s not just about how much pollution is produced at the moment,” he said. “It’s a growth and sustainability issue.”
The airline industry projects 7.2 billion passengers will take to the skies in 2035, nearly double the number this year. Total air traffic from both passenger and cargo flight will climb more than 85 percent during the same period, industry forecasts say.
But emissions could triple by 2035 under “business-as-usual” conditions, according to the UN’s International Council on Climate Change. There’s evidence that greenhouse gases from aircraft have a larger effect on the atmosphere because they are released at high altitude.
Lighter Planes Take Off
The new software, called PyFR, is based on the Python programming language. It’s a flexible, cross-platform application that makes it possible to use Python’s compact coding for high performance computing.
To show how PyFR could be used to help develop lightweight, more efficient aircraft engines, the researchers ran high-resolution GPU-accelerated simulations to measure airflow over a jet engine component known as the low-pressure turbine.
To do so, they used our CUDA Basic Linear Algebra Subroutines (cuBLAS) library and two of the world’s most powerful supercomputers — Oak Ridge National Laboratory’s Titan and Piz Daint, at the Swiss National Computing Center. Both supercomputers use our Tesla K20X accelerators — more than 18,000 on Titan and more than 5,000 on Piz Daint.
Greener Aircraft on the Way
The low-pressure turbine and its blades take energy from exhaust gases exiting the engine and use it to power the plane’s compressor, fan and other components.
Modern jet engine turbines are designed to use as few blades as possible, but this can lead to unsteady airflow patterns that reduce engine efficiency. High-resolution simulations allow engine makers to see fine-grained details that can help them get around problems like this.
“If we can simulate the airflow accurately, it tells us everything you need to optimize the design of the turbine blades,” Vincent said.
The team’s next step is to deliver the technology to industry.
“As aviation grows and connects more people around the world, greener aircraft will be essential,” Vincent said.
For an overview of our activities at Supercomputing 16, click here.