RESEARCH HIGHLIGHTS

Geothermal heat generated in the Earth’s crust is currently one of the most promising clean renewable resources for heating communities. Unlike wind turbines, geothermal plants provide a virtually limitless supply of heat regardless of weather conditions. There are, however, risks involved in the extraction process, which requires drilling deep underground, circulating fluids through hot rock layers and extracting the heated liquid back to the surface. If subsurface layers are unstable, the added pressures from the injected fluids could trigger earthquakes, posing a danger to both people and structures. Although damaging earthquakes in geothermal projects are rare, communities still need to thoroughly assess these risks before deciding if geothermal is a wise choice.

In an ongoing effort to safely and efficiently expand geothermal heating networks in the German state of North-Rhine Westphalia (NRW), a team of researchers led by Dr. Claudia Finger of the Fraunhofer Research Institute for Energy Infrastructures and Geothermal Systems (IEG) in Bochum has been developing an innovative computer modeling method that improves how we track earthquakes, assess seismic triggers, and identify geothermal hazards in urban areas. With the JUWELS supercomputer at the Jülich Supercomputing Centre (JSC), the team applied this advancement to successfully design a network of seismic monitoring stations near the city of Aachen.

Time Reverse Imaging

To assess potential geothermal drilling sites, Finger and her colleagues first needed a computer modeling method that could more precisely locate earthquakes within a limited area of the crust.

“There are many tools that can locate earthquakes robustly. It's done all over the world all the time,” Finger said. “However, when you go into a specific local region, for example for a geothermal site, and you want to look in detail, you are not only interested in the larger earthquakes that could cause harm, but you are also interested in the smaller ones. This bit of crackling in the underground can also tell you a lot of information. And to get to those, we developed a method called time reverse imaging.”

As the name implies, time reverse imaging is a numerical modeling method that relies on collected data to create an image of seismic wave propagation. By reversing the model, the researchers can trace an individual waveform back to exactly where an earthquake occurred. This, in turn, informs them about the potential for future seismic activity in specific locations.

Finger says the team was especially interested in achieving higher resolution velocity models from a particular kind of seismic wave called a Rayleigh wave. Rayleigh waves are commonly found all over the world and can be caused by many things, such as ocean waves hitting the shores. They travel along the Earth’s surface and extend downward, changing in amplitude and velocity as they move through different mediums. By applying an additional technique called beamforming—processing and filtering the summation of data across a network of sensors to estimate direction and speed of seismic waves—the researchers create a high-resolution map of nearby underground structures based on how Rayleigh waves behave. This feat of computing not only improves the ability to locate earthquakes, but also the ability to predict where future earthquakes may occur.

Cutting Through the Noise

Finger and her team’s next challenge was to design their own seismic sensor network and test the reverse imaging method in Aachen.Unfortunately, cities are notoriously difficult areas in which to measure seismic activity because vibrations from human activity create both literal and figurative noise—both environmentally and in the data. Smaller earthquakes are already harder to detect than larger ones, but they are especially obscured in a “noisier” location.

To limit this interference, the researchers again used the time reverse imaging method’s logic, but this time it wasn’t to locate earthquakes that already happened. Working backwards instead, Finger explained, they numerically calculated virtually every possible earthquake that might occur in the specified area, then tested for the optimal placement of new seismic monitoring stations based on how those waves would be expected to propagate. This way, the amount of noise could be limited while the amount of seismic data collected could be maximized. Thanks to JUWELS, this intensive set of calculations revealed an excellent map to work from in their design plans.

However, the optimal locations identified were not all feasible in a practical sense. “It's an urban area, so there are a lot of people, a lot of industry, a lot of things going on,” Finger said. “On the other hand, we also do not have these wide-open fields where we can deploy our stations either. And they are a bit expensive, so just putting them on the side of the road is risky.”

For the project to move forward, it was going to require some help from local residents.

Valuable Public Participation

 In order to place sensors as closely to optimal locations as possible, the researchers engaged with the local communities of Aachen and Eschweiler, a small town close to Aachen. They asked permission to set up sensors on private property, offering citizens a close-up look at how public science resources are used to benefit their lives. In return, the team gained a supply of exclusive experimental data that they could not have otherwise accessed.

“We came up with the idea to just involve the public from the beginning,” Finger explained. “We said, ‘Hey, we can put it in your garden. It's a small thing, it will not disturb you.’ Then we also get to raise awareness of what we're doing and why it is important to get them on board, which in the end it benefits both sides.”

“Dr. Finger’s project is an excellent example how publicly funded research infrastructures, like the supercomputer JUWELS, enable to solve research challenges with immediate benefit for society,” said Prof. Dr. Dr. Thomas Lippert, Director of the Jülich Supercomputing Centre. “Involving the local community, as done by her team, is an exciting element that closes this circle of research for and engaging with citizens.”

Through this combination of theory and experiment, the team was also able to improve its beamforming computational code to be more accurate in measuring risks connected to so-called sedimentary basins in the sub-surface—large sections of sedimentary rock that are looser and, as a result, have the potential to allow larger, more violent seismic waves close to the surface, endangering people and structures.

For Finger and her team, results like these are only possible with a combination of HPC-enabled scientific innovation and real community engagement.

“When we installed these sensors, I had the chance to talk with individual people about their concerns and questions about this kind of work,” she said. “They want to know the risks to their homes, and whether geothermal energy is the future in their community. I then get the opportunity to share learning resources about the supercomputer we use, point them to publications about related research we’ve been doing, and actually show them data that was recorded in their gardens. This engagement makes science more approachable for my community.”

-Sarah Waldrip