The University’s Laboratory for Laser Energetics (LLE) is home to Conesus, one of the world’s most energy efficient supercomputers. With the capacity to create and store clean energy from nuclear fusion and building on last year’s ignition milestone, researchers believe that energy efficiency goals are more possible than ever before.
How is fusion created?
Dr. Chris Deeney, director of LLE, explains their process to create fusion in relatively simple terms. Researchers at LLE use lasers to hit a cryogenically cooled pellet of deuterium tritium fuel (hydrogen isotopes) that is approximately a millimeter wide for only one billionth of a second. This compresses the fuel pellet and creates an intense pressure and temperature, releasing energy when the deuterium and tritium fuse to helium and a neutron. Fusion is a process that occurs naturally in stars like our sun, with Deeney saying fusion experiments on Earth are “creating stars in the laboratory.” Due to its potential to become an incredibly important sustainable energy alternative, fusion companies around the world are accelerating the pace of development. Since fusion does not directly release any greenhouse gasses, doesn’t create long-term radioactive waste, and it mainly uses elements already abundant in nature, it could be a great addition to renewable energy sources to address energy demand.
Fusion experiments are conducted at LLE about once a month, with scientists only getting to shoot the laser about five times. While they receive a lot of data from these experimental days, it is not completely sufficient for their research needs. “That’s where the computers come in,” says Deeney. LLE has computer codes that can simulate these fusion experiments and use those results to inform their next attempts. The longer they use the computer and update the code, the more accurate the simulations can become.
What is Conesus?
Why is a supercomputer important?
These code calculations are very complicated, taking weeks to run. With Conesus, projects that could have taken up to 30 weeks can now be completed in only a few days. Deeney notes that this increased computational power also allows them to add more physics to these simulations, making them more accurate. There’s also now the possibility to apply machine learning or artificial intelligence approaches which require large amounts of data. Running more simulations will help build up LLE’s database faster, improving their strategies.
In addition to Conesus, LLE has acquired a new advanced storage system named Erie from IBM thanks in part to state funding secured by NYSERDA and Empire State Development. These experiments create large amounts of data and as their operations grow, Erie will be essential for providing increased storage capacity.
Energy Efficiency
Conesus is notably a highly energy efficient supercomputer, winning a spot on the Green500 list, which is a subset of the TOP500 list of powerful supercomputers. One factor that made Conesus reach higher energy efficiency was water cooling, rather than the standard air cooling. Water cooling allows heat to be extracted more efficiently, making a big difference on power needs.
Computers are the biggest consumer of electricity at national labs, Deeney explained, so there’s been a national effort to improve computer energy efficiency. This means LLE will be able to lower its greenhouse gas emissions, as running and maintaining their supercomputer will require less electricity than in the past.
Deeney noted that this new system could not have been brought together without leadership of Valeri Goncharov and William Scullin at LLE. “Fusion and supercomputers are all team efforts” said Deeney
Written by Sarah Woodams ‘24(T5)
Photo by J. Adam Fenster / University of Rochester