AsianScientist (Jul. 5, 2023)
By Marinel Mamac and Jihan Al-Shdifat
When it debuted in 2020, the supercomputer Fugaku was hailed as the centerpiece of Japan’s Society 5.0—a vision of a country able to solve social problems and advance its economy with the help of digital technology. It was, after all, the world’s fastest supercomputer at the time, a title it held on the biannual TOP500 list from its launch until June 2022.
“The main mission of Fugaku is to attain sustainability goals in the areas where it is involved,” said Professor Satoshi Matsuoka, director of the RIKEN Center for Computational Science and part of the team behind Fugaku, in an interview with Supercomputing Asia.
Achieving 442 petaFLOPS of computational power, Japan’s fastest supercomputer was developed by scientists at RIKEN with an application-first philosophy. That is, it wasn’t just about achieving computational excellence for its own sake—the machine was built to solve the biggest crises of our time. “Many of these crises pertain to carbon neutrality,” Matsuoka noted.
More than just a buzzword, decarbonization has become a crucial goal among researchers in Japan—a country that is among the world’s top carbon emitters. By the year 2030, Japan aims to reduce carbon emissions by 46 percent compared to 2013 baseline levels. With today’s generation of high performance computing (HPC) hardware, software and talent, Japan is leading Asian countries in achieving this goal.
Tacking the problem of carbon
In the half-hour or so that you spend sifting through this magazine, approximately 1.9 million metric tons of carbon dioxide (CO2) were emitted globally as a result of burning fossil fuel. Of the 50 billion tons of CO2 and CO2-equivalent greenhouse gases emitted each year, 73.2 percent comes from the energy sector, 18.4 percent from agriculture and 5.2 percent from industry. The remaining 3.2 percent comes from landfills and wastewater.
Of course, the earth has its own ways of ensuring a balance: Up to 83,000 metric tons of carbon per square kilometer will be sequestered by seagrass meadows, while forests take up around 30,000 metric tons of carbon per square kilometer.
But still, some 40 percent of emissions will make their way to the atmosphere, while 30 percent will be absorbed by seawater, causing ocean acidification. And as carbon emissions continue to rise and we lose our forests to rapid urbanization, wildfires, mining, unsustainable agriculture and rising sea levels, decarbonization innovation has become increasingly crucial.
Tackling the world’s carbon problem entails two sides of the same coin. The first involves reducing greenhouse gas emissions—from using renewable energy sources to controlling carbon emissions in agriculture. This also includes ensuring better energy efficiency across different industries. The second involves finding ways to improve the absorption of carbon from the atmosphere, either by capturing emissions directly or enhancing the natural carbon storage of our forests and seagrass meadows.
It’s a tall order—but it’s one that scientists at RIKEN and the rest of the world are taking on with the power of HPC.
Reducing carbon emissions
Primary among Fugaku’s decarbonization research goals is developing a strong pipeline of renewable energy. Matsuoka explained that Japan’s biggest hope for carbon neutrality lies in wind and solar energy.
In line with this, the country aims to generate 10 gigawatts of offshore wind power by 2030. To make this possible, the country has been investing in humongous offshore wind farms, with propeller blades as high as 200 meters.
“Each blade is like a skyscraper,” shared Matsuoka. “These things are so large that the blades sit above the clouds and may actually affect the weather. This makes them very difficult to design.”
To make these offshore wind farms work, scientists have to consider more than just the physics of converting wind energy into electricity. They are using Fugaku’s computing power to run simulations based on extreme weather conditions, blade materials and overall design.
As for solar energy, Matsuoka said that Japanese material science teams have been studying how to make safer and more efficient photovoltaic cells locally and, in the process, would need to investigate some 20 million possible substances. “That number is just too huge to run a [physical] experiment,” he pointed out.
With Fugaku and artificial intelligence systems designed for solar energy research, scientists can run billions of simulations on all these possible substances, filtering through these compounds much more efficiently. “One of the substances discovered by our team has an energy conversion rate of up to almost 25 percent. This is very promising,” Matsuoka said. From there, the next step is to translate these photovoltaic cell simulations to real life and then conduct further research, he said.
Aside from energy generation, Matsuoka shared that food and agriculture is another crucial area of research being conducted with Fugaku. Food production generates lots of carbon, he pointed out, with cows being the largest producers of greenhouse gases in the sector. Today, Japanese scientists are trying to develop a breed of cows that produce less methane, while others are investigating better and more efficient ways to produce food.
“What’s interesting for me is that the technologies these teams are using are very similar to technologies we have already developed for other areas of biology, like human medicine,” shared Matsuoka. “Some of the infrastructure we have built to accelerate drug discovery can be translated into food production, because underneath they’re using various genomic
and proteomic technologies.”
Another area that Japanese scientists are looking into is maritime transport. Today’s ships are the backbone of global trade, but they also consume a significant amount of the world’s total energy— prompting scientists at RIKEN to study how we could make more efficient ships.
Matsuoka explains that designing these ships has always been a huge challenge. In conventional development, maritime design requires scale models. Huge pools are built to test small ship prototypes, with some pools being hundreds of meters in length.
“For the first time, we have the capacity to assist scientists in the maritime design process, eliminating the need for these huge pools,” Matsuoka shared.
Scientists at the University of Tokyo have begun using Fugaku to conduct pool simulations, taking into account thousands of moving factors like the viscosity of water, the design of the cargo ship and the screws that will propel it. This allows them to boost ships’ energy efficiency by 10–15 percent, which Matsuoka pointed out can make a huge impact globally.
“There are many areas—from power to various industries like food and materials—where Fugaku is contributing,” said Matsuoka. “We haven’t done a cumulative assessment of just how much potential Fugaku has allowed us to tap into. Maybe that’s something we should do, collectively, so we can continue to significantly reduce Japan’s overall carbon footprint.”
After reducing carbon emissions, the other side of the global decarbonization effort is capturing carbon already in the environment, a process called carbon sequestration.
Scientists around the world have begun researching how to trap CO2 underground, an effort that Fugaku and other supercomputers have assisted in. For this method of carbon sequestration to work, scientists need HPC to simulate millions of scenarios to understand the best way to inject CO2 into the ground—such as the optimal place and method—and afterwards, to prevent it from escaping.
Another path for carbon sequestration is enhancing the earth’s natural systems for balancing CO2 levels. Researchers at the RIKEN Plant Science Center are looking into using plants both for producing energy and absorbing more CO2 from the atmosphere using Fugaku.
These efforts are conducted in parallel with research elsewhere in the world. Blue Waters, a petascale supercomputer operated by the University of Illinois, has worked to improve tree mapping efforts for non-forest trees, which function as significant carbon sinks but are less studied than trees found in forests. Meanwhile, the Pawsey Supercomputing Centre in Australia is looking into ways to turn CO2 itself into fuel.
A shared goal
For Matsuoka, the race to net zero is not a race among different countries—it’s one we are collectively competing in against time. This is why, as part of Fugaku’s application-first philosophy, the RIKEN center provides compute cycles to climate and decarbonization projects to be rolled out across Asia.
For instance, one initiative Fugaku will soon be used for is a microalgae-based carbon capture and utilization project for Indonesia under the Japanese government’s Science and Technology Research Partnership for Sustainable Development program. Led by Tokyo Institute of Technology’s Dr Muhammad Aziz, the project aims to capture CO2 and use it as a sustainable resource with the use of microalgae. “In addition to its scientific contribution,” Aziz told Supercomputing Asia, “This study is also expected to help in the reduction of CO2 and nitrogen oxides from power plants.”
Aside from providing access to HPC resources, the RIKEN team also provides training and education to those interested in conducting HPC-enabled research, depending on skill level and project complexity. Matsuoka described this process as one where newbies and seasoned veterans “work their way up the mountain,” as Fugaku takes its name from the alternative name for Mount Fuji—Japan’s highest peak, but one that Matsuoka says is an easy mountain to climb.
This way, the team behind Fugaku is providing HPC resources for all interested researchers, from beginners to real experts, who want to address our world’s biggest crisis.
“At the end of the day, it doesn’t really
matter who solves the problem of carbon,” Matsuoka said. “As long as
the problem is solved.”
This article was first published in the print version of Supercomputing Asia, January 2023.
Click here to subscribe to Asian Scientist Magazine in print.
Copyright: Asian Scientist Magazine. Illustration: Shelly Liew/Supercomputing Asia
Disclaimer: This article does not necessarily reflect the views of AsianScientist or its staff.