Making geothermal energy safer through simulation

After the earth trembled in the St. Gallen Sittertobel in 2013, the test drillin
After the earth trembled in the St. Gallen Sittertobel in 2013, the test drillings for the St. Gallen geothermal energy project were stopped. (Image: Stadt St. Gallen / St. Galler Stadtwerke)

Researchers from the Swiss Seismological Service SED and ETH Zurich are working with the Swiss National Supercomputing Centre CSCS to develop a way of utilising geothermal energy safely with the help of supercomputers.

According to Switzerland’s Energy Strategy 2050, the plan is for deep geothermal energy to contribute to the expansion of renewable energies in the country. However, this ambition presents special challenges. Although it is relatively easy to utilise geothermal energy on the volcanically active island of Iceland, for example, it is often much more difficult - and risky - on the continents.

In Switzerland, engineers must drill to a depth of between 4 and 5 kilometres to reach regions of the Earth’s crust that are hot enough to heat water to the required temperature of 160░C to 180░C. This water must flow through the hot rock via a borehole before being pumped back up to the surface. However, one problem is that the rock is not very permeable at these depths. "We need a permeability of at least 10 millidarcies, but typically we only find a thousandth of that value at depths of 4 to 5 kilometres," says Thomas Driesner, professor at the Institute of Geochemistry and Petrology at ETH Zurich.

To make the rock more permeable, pressurised water is pumped into the underground. The water pressure counteracts friction on fractures, allowing fracture surfaces to slip and thus release stress. This hydraulic stimulation widens the fractures and makes them more permeable, so the water is able to circulate within the hot crust. Such fractures in the Earth’s crust are the result of tectonic stress, which in Switzerland is caused by the Adriatic plate moving north and pushing against the Eurasian plate.

Water injection causes earthquakes

The disadvantage of hydraulic stimulation is that the resulting movements cause shaking. This should typically be so slight as to be only barely perceptible to people, if at all. But unfortunately, this was not the case with the geothermal projects in St. Gallen in 2013 and Basel in 2006. In both cities, earthquakes were recorded with a magnitude higher than 3, and in Basel local residents not only noticed the resulting shaking, they also heard a loud bang.

"Prior to that, in Basel a total of about 11,000 cubic metres of water had been pumped into the borehole, increasing the pressure," says Driesner. Statistical surveys had been used to define magnitudes of 2.4 and 2.9 as two limit values for the maximum permissible strength of generated earthquakes, and so the water supply had been stopped in time. Nevertheless, stronger earthquakes occurred after a delay. This ultimately undermined public acceptance of geothermal energy in Switzerland.

Since the incidents in Basel and St. Gallen, it has become clear that the concept of simply stopping at a specific threshold is not sufficient. The Swiss Seismological Service SED has therefore been looking for an alternative way to improve predictability and safety. Their new plan is to use an Advanced Traffic Light System based on rock physics that will predict, almost in real time, whether perceptible earthquakes can be expected as hydraulic stimulation continues. To achieve this, the SED began developing software to analyse the response of the underground during hydraulic stimulation.

However, it soon became clear that this would require optimised software and High Performance Computing (HPC). So the Platform for Advanced Scientific Computing PASC was used to launch in 2017 a project entitled "Forecasting and Assessing Seismicity and Thermal Evolution in geothermal Reservoirs" FASTER, led by Driesner and including researchers from the SED, ETH Zurich, UniversitÓ della Svizzera italiana USI, and software engineers from the CSCS.

Simulation to investigate the underground

In the initial stage of hydraulic stimulation, the quantities of water are small and the shaking is therefore minimal. Geophones - that measure ground shaking - indicate where and at what distance the tremors occur around the borehole. These signals already contain sufficient information for simulations on the supercomputer to estimate the probability of perceptible earthquakes as stimulation continues.

The aim of FASTER is to feed this data in real time into the CSCS supercomputer "Piz Daint", which will then run through millions of potential scenarios: the possible number of fractures, their type and orientation, and how much friction and stress they can sustain. "Of these scenarios, 99.99 percent are completely unrealistic, but we don’t know which ones in advance," says Driesner. In a very short span of time, "Piz Daint" can analyse which of the scenarios best reflects the underground. This allows the situation to be assessed within minutes so a decision can be made on how to proceed, Driesner explains.

But there is still one problem: the researchers lack a real-world test opportunity for their system. As Driesner emphasises, it is important to eliminate occasional errors in the measurements and comply with a specific data format before performing the calculations on the supercomputer. The researchers had an opportunity to test the communication between the measuring equipment and the computer during a stimulation in Iceland last year, but only sparse seismic activity occurred. The researchers now hope to be able to test their tool at ETH Zurich’s Bedretto Lab in late summer.

In this underground geothermal laboratory, they plan to hydraulically stimulate a volume of rock between two boreholes, 100 metres apart, at a depth of approximately 1,200 metres. The aim of these experiments is to investigate whether controlled stimulation can create sufficient permeability and also keep the strength of the induced earthquakes under control. "There’s a fairly clear correlation - the more water we pump underground, the stronger the quake," says Driesner. "So, it’s a question of assessing how much water to inject in order to stay on the safe side while also increasing the permeability of the rock."

Boosting acceptance of geothermal energy

The researchers believe that their tool may not only help to make geothermal energy a safer renewable option but also could also be of use wherever there is a risk of artificially induced earthquakes - such as in underground mining or the underground storage of CO2. They also hope to use the knowledge gained to establish best practices in geothermal engineering that emphasises the appropriate planning for, and successful implementation of, geothermal plants.

To this end, and to pave the way for the eventual routine installation of deep geothermal energy plants as originally envisaged in Energy Strategy 2050, Driesner says that there will be a need for pilot plants to support continued application-oriented fundamental research.

Simone Ulmer