Supercharged by the high-performance computing power of Summit, this public-private partnership’s fundamental research will help produce a new, hyper-efficient engine for propulsion and power generation
Revisiting an engine concept first proposed in the 1950s, researchers at the University of Michigan (UM) are conducting trailblazing research that may finally unlock its potential for ultra-high-efficiency propulsion and power generation.
UM professor of aerospace engineering Venkat Raman has tapped the nation’s most powerful supercomputer for open science, Summit, to execute combustion simulations with unprecedented fidelity and speed for this new engine. Summit is located at the Oak Ridge Leadership Computing Facility (OLCF), a US Department of Energy (DOE) Office of Science User Facility at Oak Ridge National Laboratory. The groundbreaking simulations are giving engineers at General Electric (GE) Research new insights they believe will enable them to forge the breakthroughs required to commercialize this long-sought-after technology: a rotating detonation engine (RDE).
Many aerospace companies and governments around the world are actively pursuing RDE technology as well, but access to the processing power of the IBM AC922 Summit has proved to be a critical advantage for his project, Raman said.
“To me, Summit is a step change. Its compute power is something I did not expect. Even the biggest calculations that we thought we were going to run would fit on 20 nodes of Summit—and Summit is over 4,000 nodes. So we really had to increase our ambition once we got access to the machine,” Raman said. “I think that sort of step change requires you to rethink how you do your simulations and what simulations can actually be done. We changed our algorithmic frameworks, we changed our codes, we even changed how we answer the questions.”
This innovative work is the result of a collaboration among government, academia, and industry. The primary funding for Raman’s research comes from the DOE’s Office of Fossil Energy through the University Turbine Systems Research Program, which is overseen by the National Energy Technology Laboratory (NETL) and its Advanced Turbines Program. The actual computing time on Summit was allocated by the DOE Office of Advanced Scientific Computing Research’s Leadership Computing Challenge (ALCC). GE, an industry leader in both propulsion and power-generating turbine technology, is collaborating with the goal of accelerating its own RDE projects.
Together, these institutions seek to put practical applications of RDE technology on a fast track to implementation, promising higher fuel efficiency than current turbine-engine technology. This could mean the near-future development of jet airplanes and power generators that leave a much smaller carbon footprint.
What are rotating detonation engines?
The concept for RDEs was pioneered in the 1950s by the late James Arthur Nicholls, a professor in UM’s Department of Aeronautical (now Aerospace) Engineering. Like many innovations in engineering, his idea was the result of asking an unorthodox question. Noting how malfunctioning rocket engines tended to explode, it made him wonder: What if such explosions were used in an engine to propel rockets?
A conventional rocket engine uses compressors to send a pressurized mixture of fuel and oxygen into a combustor where the mixture ignites and burns via deflagration—essentially the same combustion process used by car engines or jet turbines. This deflagration releases energy, which can be used to drive a mechanical device (such as a turbine) or to thrust a rocket upward. An RDE, on the other hand, burns its fuel by detonation rather than deflagration—combustion occurs in a wave front that is led by a sustained shockwave that continuously rotates within a cylindrical combustor’s inner and outer walls.
In the RDE combustion process, the detonation wave builds its own pressure rather than losing pressure as occurs in conventional compressor-equipped engines. The result is a much higher degree of efficiency in pulling energy out of the fuel since it, in effect, acts as its own compressor even as it’s releasing energy. While a detonation engine’s burn rate is much faster than that of conventional turbine engines, its combustion chamber is also much smaller, resulting in higher pressure levels—and even more efficiency in extracting the energy.
Such efficiency gains are especially appealing for electricity production—a keen point of interest for NETL, which focuses on transitioning technology from the laboratory to industry.
“The application we’re looking at here is in power generation with a 3 to 5 percentage-point improvement in overall cycle efficiency. This is huge in the gas turbine world, where they’re usually looking at a tenth of a percent of improvement in efficiency. Ultimately, that boils down to lowering your energy bill,” said Don Ferguson, a research engineer with the Thermal Sciences group at NETL’s Research & Innovation Center. “If we can produce more power for a given amount of fuel, then ultimately you reduce the cost of electricity as well as reduce the amount of carbon that’s generated at the power plant. We’re extracting more energy out of the fuel, in this case maybe natural gas or some coal-derived syngas.”
However, this pressure-gain combustion process is very difficult to study and develop, which is one reason why RDE technology hasn’t been fully realized since Nicholls’ initial prototypes. Conventional turbine engines have been much simpler to refine—increasing their efficiency typically meant raising the gas temperature going into the turbine or raising the compressor’s pressurization—so that’s where most R&D money went for the past 50-odd years. But now we’re reaching the limits of materials science to increase temperatures or pressure in these engines, which has renewed interest in RDE technology and its potential to burn a variety of different fuels. However, its extreme variations in temperature (between 300 and 3,000 kelvin) and the speed of its shockwaves (2 kilometers per second) make experiments extremely difficult.
“Because the physics is so complex, the operating environment so harsh, there are limited measurements that we can make. We can measure macroscopic things like pressure and temperature, but it’s hard to get in there with lasers and stuff like that to really understand the physics of what’s going on,” said Peter Strakey, a research scientist with NETL’s Thermal Sciences group. “We have to understand this optimization process—how do you redesign the injector and the fuel manifold and the air manifold to actually achieve a pressure gain? The simulations are critical to understanding the physics and being able to kind of see inside these devices and understand what’s going on.”
Understanding all of the physics at play during an RDE detonation wave is critical to making the technology safe for the commercial uses GE is pursuing. For example, in order to replace an aircraft engine with a detonation engine, you must take into account all of the plane’s different operating conditions—taxiing, taking off, cruising—to ensure the detonation engine’s efficiency remains stable. And to do that, the first step is to understand what happens to the key physics inside the engine when those conditions change.
“So many of the techniques we have for conventional jet engines and gas turbines don’t work in these kinds of extreme environments. So simulations are the only way to go. There’s no way around it,” Raman said. “However, simulating the complicated physics in RDEs is very challenging. These simulations have as much physics as the most complicated problem you can think of—fluid mechanics, shockwaves, chemical reactions, a heat transfer to the wall.”
Fortunately, Raman did find a way around those obstacles to produce the most insightful combustion simulations to date.
Gaining new insights into RDE
Raman and his team of students knew that Summit would provide faster simulation results—after all, that’s why they applied for the ALCC allocation—but they had no idea it would also give them the opportunity to advance the state of the art for computational fluid dynamics tools.
At first, the team got off to a slow start, putting only about 30 percent of their simulations code—based on the open-source modeling software OpenFOAM—onto Summit’s GPUs. This approach did not result in much of a speedup. But, with guidance from OLCF staff, the students completely rewrote OpenFOAM to take full advantage of Summit’s GPU-powered architecture—now, 95 percent of the code runs on GPUs. Furthermore, the team recognized the opportunity to leverage the advantages GPUs have for artificial intelligence algorithms and consequently trained neural networks to speed up the modeling program even further—a first in the field. All of these improvements led to GPU performance reaching near theoretical limits, restricted only by the communication bandwidth for transferring data between the CPU and GPU.
“This allows us to get—in some extreme cases—about 1,000-times-faster simulations compared to if you just did conventional calculations,” Raman said. “More than even speedup, you can also answer different questions. It’s not an incremental change.”
GE’s Venkat Tangirala, a principal engineer and the RDE program manager at GE Research, has been studying pressure-gain combustion and RDE technology for some 20 years, yet he’s excited and impressed by the results he’s seen so far from Raman’s work on Summit.
“Venkat Raman’s advanced code for Summit stands head and shoulders above anybody else’s. He’s delivering stunning results,” Tangirala said. “It’s not just ‘show and tell.’ He can do the high-fidelity simulation of the detonation structure and at the same time do it faster. I can’t emphasize enough: his work will speed up research by an order of magnitude when the code is made available so that we can all be using it in the technology development process.”
For example, simulations of traditional aircraft engines may represent 100 or 200 milliseconds in real time, but RDEs have typically required about a full second of simulated time to get meaningful insights. “Such calculations would take months on a CPU-only machine. With Summit, you’ll finish the simulations in about a day,” Raman said. Summit’s GPUs enable parallel simulations; for example, if you run the full-scale RDE simulation using100 million grid points on 20 nodes, you can run 200 such calculations on Summit at the same time, utilizing 4,000 nodes, he said.
“You cannot just use conventional techniques to solve these problems—you really have to believe that GPUs are the way to go and reset everything you know about solving fluid mechanics. That’s the only way to get to those execution times,” Raman said. “So that’s why I think a machine such as Summit or the ones that are coming up, all the next-generation supercomputers, will all rely on GPU-type parallelism in one way or another. We have been able to use this hardware effectively by recognizing that a big data/machine learning software stack can be repurposed for scientific computing.”
Bringing RDE to market
Raman’s speedup of RDE simulations on Summit will mean faster development of the real-world technology. Tangirala believes the data that Raman’s team has been gathering will be integral to GE’s ongoing development of different applications, from stationary gas power generators to hypersonic aircraft engines.
“Venkat Raman’s code is the fastest in the field out there,” Tangirala said. “The turnaround time we are getting on Summit, combined with the quality of these high-fidelity simulations, is as good as doing physical tests. We never thought this would be possible.”
Tangirala foresees three main technical challenges to overcome for RDE’s commercialization: (1) quantifying the pressure gain and figuring out how to optimize the design for increased pressure gain; (2) designing cooling methodologies that will make sure the engine materials can withstand RDE’s high temperatures; and (3) ensuring that the overall engine design has enough thermal efficiency for a long lifecycle.
“Those are the areas where we think Dr. Raman’s research will really help us,” Tangirala said. “We have the rotating detonation technology and test rigs here for experiments. And now we also have his computations going hand in hand to help us understand what our tests are telling us. I really don’t know how many research institutions are set up to do that, but that’s how we’re making tremendous progress.”
Beyond higher efficiency, Raman sees another advantage to RDE technology: new options for machine design. For example, detonation engines would not necessarily need to be located under jetliner wings like conventional turbines—they could be placed in more compact locations or even wrapped around the fuselage.
“You can come up with very unconventional designs that so far are not possible. That, to me, is a game-changer. All of this research into RDEs came because we thought we’ll get a higher efficiency, but I think this has really opened up the design space,” Raman said. “Jet engines and stationary gas engines, which generate electric power, are nominally the same kind of devices, but they have all different kinds of physics. But in the case of RDE, you can deal with one physics and target many different applications.”
Tangirala predicts that we’ll see RDE-powered rockets soon, followed soon after by aircraft and power-generators. Raman is equally optimistic.
“I think we’re at the point where the first machines that will use this technology are being designed right now, so it’s very possible that we’ll see them in use within the next few years,” Raman said.
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 https://energy.gov/science.