Simulations could help head off billion-dollar problem

Gas-fired turbines keep electric generators humming around the world, from power plants and factories to hospitals and universities. But the combustion process that fuels these engines comes with a billion-dollar-a-year problem: Under the wrong conditions, the resulting heat and noise can cause tremors and vibrations known as thermoacoustic oscillations, which can damage parts, shut down operations and even shake the machinery apart.

For the turbine industry, it’s a complex problem that can be difficult to predict. Large turbines can take years to design and build, and the flaws that lead to the potentially damaging oscillations might not show up until all the components are made and assembled at the final stage, forcing costly do-overs.

Simulations performed on Oak Ridge National Laboratory’s Summit supercomputer promise to help solve that problem by enabling engineers to understand and predict those oscillations early in the design process. This computational approach had been tried before, but previous efforts stalled for lack of the extraordinary computing power required to model the complicated flow of gases during combustion.

An industrial gas turbine

Gas turbines power industries around the world, but vibrations from heat and noise can shake the systems apart under the wrong conditions. Simulations performed on Oak Ridge National Laboratory’s Summit supercomputer could help head off that problem. Credit: Solar Turbines

Engineers for Solar Turbines, a subsidiary of Caterpillar, and software partner Cascade Technologies modeled the turbulent flow inside a 14-burner gas turbine combustion chamber for the first time – one of the most detailed simulations of its kind to date. The simulations successfully predicted the oscillations, which could transform the design process.

Solar Turbines manufactures a widely used family of mid-sized industrial gas turbines, some as small as a single megawatt in capacity and others as large as 23 megawatts. The company’s extensive line of products powers a variety of industries, from natural gas and crude oil production, processing, and pipeline transmission, to generating heat and electricity for factories, power plants, and rescue and relief operations.

The research team’s findings offer a potential guide for improving current gas turbines as well as for designing new models for alternative, carbon-free fuels such as hydrogen, said Yonduck Sung, a senior principal engineer for Solar Turbines who led the study.

“These insights hold paramount importance for our efforts to continue delivering reliable gas turbines without compromising emission standards and performance,” Sung said. “Simulations like these can speed up important discoveries by replacing expensive physical tests and reducing time in the design cycle.”

A risky ratio

Industrial gas turbines rely on a delicately balanced mix of fuel and air to produce maximum energy from the minimum amount of fuel. Lean mixtures like those used in Solar Turbines products lower the fuel-to-air ratio even further to reduce toxic gases such as nitrogen oxides that contribute to air pollution. Poor mixing before combustion can raise the flame temperature, increasing nitrogen oxide emissions.

Combustion of these lean mixtures poses a series of technical challenges that call for precise control of fuel injection and air flows along with other variables such as injection location, direction, and velocity. Fluctuations in any of these variables can lead to noisy and potentially damaging thermoacoustic oscillations.

The computational power of Oak Ridge National Laboratory’s Summit supercomputer enabled Solar Turbines engineers to model a 14-burner gas combustion chamber. Credit: Solar Turbines

“At the extreme, these instabilities can shake a system apart, and they often don’t appear until late in the design cycle when all the components are built and assembled, and a physical test of the full engine is performed,” Sung said. “At that point, we may have to start all over.”

Solar Turbines’ engineers sought a way to understand and reliably predict thermoacoustic oscillations early in the design phase, which can take as long as 3 years, without burning time and money on a cycle of repeated physical tests and redesigns. The complicated variables of the combustion process and the intricate relationships between turbulence, chemical reaction and thermoacoustics made that mission easier said than done.

“We need to be able to spot the problems before they start and before we cut metal,” Sung said “But understanding and predicting these combustion instabilities using high-fidelity numerical models requires very large computational resources that far exceeded our in-house computing systems. We needed more computing power.”


Capturing the complexities

Researchers from Solar Turbines and Cascade Technologies teamed up to approach the Oak Ridge Leadership Computing Facility, home to Summit, the US Department of Energy’s 200-petaflop IBM AC922 supercomputing system. The team received an allocation of computing time on Summit through a director’s discretionary grant via the industrial partnership program Accelerating Competitiveness through Computational Excellence, or ACCEL.

Solar Turbines’ engineers could model only one burner unit using the company’s in-house computing system. That limited attempt couldn’t capture the distinctive noise and vibrations of the turbines, known as the acoustic signature, or the complicated multiphysics of combustion-driven thermoacoustics, Sung explained.

The power of Summit enabled simulation of a full combustion chamber, complete with a full annular combustion section and 14 operating burners. The team used charLES, a high-fidelity computational fluid dynamics code developed by Cascade, to perform large-eddy simulations, which capture turbulence, and study the interplay between the combustion and engine pressure.

“This kind of problem couldn’t be run anywhere but on Summit,” said Sanjeeb Bose, Cascade’s chief technical officer. “These large-eddy simulations would have taken months to run on a conventional computer. Summit’s world-class speeds shrank that time to a matter of days.”

The combination of Summit and charLES helped capture the complex physics and geometric details needed to understand the relationships between fuel mixing, flames, and oscillations.

“These were showpiece simulations,” said Lee Shunn, a senior research scientist for Cascade who helped write the code. “Using Summit and charLES, we were able to capture everything from the big picture down to the microscopic level and make a real impact on the problem of combustion instability.

“We demonstrated it’s possible to simulate a real industrial gas turbine combustion system in full geometric detail. In doing so, we were able to predict combustion-induced oscillation in our test cases for the first time.”

The results are changing Solar Turbines’ approach to designing turbine systems, Sung said.

“Summit’s fast turnaround showed us we can produce results in an actionable time scale, shortening the design-cycle time for our products and reducing costs associated with physical testing and redesign,” he said. “It can also help to diagnose problems in the field quickly for fast fixes. This is an important first step in developing a predictive tool that can help us anticipate oscillations earlier in the design and development process for future gas turbine products.”

The success with Summit led the company to jump into high-speed simulations for research and development. Solar Turbines has increased its in-house resources for high-performance computing by 230 percent with investments in hybrid computer architectures similar to Summit’s.

Next steps for the research team include refining their predictive tools to zero in on the warning signs of oscillations and adapting the model for turbines powered by alternative fuels such as hydrogen or that can run on more than one kind of fuel. They hope to publish their research in the near future.

“Hydrogen has completely different combustion characteristics, such as higher flame temperatures and lower density, than natural gas, the most commonly used fuel for industrial turbines,” said Ferenc Pankotai, manager of combustion engineering and additive manufacturing for Solar Turbines. “So, understanding the combustion instabilities of hydrogen holds paramount importance for designing reliable turbines without compromising performance and emissions. We expect a 10 percent increase in fuel flexibility for one product line could eliminate an additional 3,800 metric tons of carbon dioxide emissions per turbine per year.”

Such insights offer a keen advantage in a highly competitive industry.

“We continually strive to stay on the leading edge of development in fuel flexibility, and these kinds of tools will be invaluable for that goal,” said John Mason, director of gas turbine technology and new product development. “Thanks to Summit, we now know how to perform computations that once seemed impractical or impossible.”

This research was supported by the DOE’s Office of Science. The OLCF is a DOE Office of Science user facility.

UT-Battelle LLC manages Oak Ridge National Laboratory for DOE’s Office of Science, the single largest supporter of basic research in the physical sciences in the United States. DOE’s Office of Science is working to address some of the most pressing challenges of our time. For more information, visit

This document was prepared in collaboration with Solar Turbines, as a result of the use of facilities of the U.S. Department of Energy (DOE) which are managed by UT-Battelle, LLC. Neither UT-Battelle, LLC, DOE, or the U.S. Government, nor any person acting on their behalf: (a) makes any warranty or representation, express or implied, with respect to the information contained in this document; or (b) assumes any liabilities with respect to the use of, or damages resulting from the use of any information contained in the document.