Safety is our number one priority. FEAR aims at inducing small, non-damaging events up to about magnitude 1, corresponding to earthquakes with rupture lengths no larger than a few tens of meters. Inducing earthquakes in the magnitude 1 range is a challenging goal for the project, as events induced so far in other experiments in the BedrettoLab (e.g., the VALTER stimulations) were several magnitude units smaller. Inducing large enough events (in the magnitude 1 range) while also limiting larger ruptures requires actively  manipulating – or preconditioning - stresses acting on the fault, one of the major objectives of FEAR.

The chance of inducing a larger and potentially felt or even damaging earthquake is small. The tunnel depth of 1 km is shallower than the typical seismogenic depth of 5-10km for larger earthquakes in the Alps. At such shallow depths, the maximum shear stress/potential energy a fault can store, and hence the earthquake size it can produce, is limited. Nevertheless, the risk is not zero. Assessing this risk, and adopting mitigating measures to minimize this risk and comply with the strict safety thresholds we have defined, is essential.

Safety of the FEAR experiments is composed of multiple layers that act in combination: 

• The BedrettoLab follows the established procedures and safety measures outlined in the BULGG risk study, which defines strict and conservative safety thresholds. For the FEAR experiments, a dedicated risk and mitigation study will be conducted and externally reviewed before the start of the major fault stimulation experiment. This risk study will include probability estimates for (i) shaking felt outside the tunnel, (ii) slight damage outside the tunnel, (iii) severe damage outside the tunnel.
• We are starting with smaller scale and limited experiments that are inherently much safer to learn from for modelling the response of the faults. 
•  In the design of the experiment, we strive to limit the ruptures by limiting them as much as feasible through stress barriers (Experiments II and III). The patches of the target fault adjacent to the Bedretto tunnel will have lower pore pressures (due to tunnel-induced pressure depletion or active draining) and would thus be less critically stressed and inhibit further rupture. 
• We will monitor the target fault with an extensive monitoring system installed at very close distance to the fault and will thus be able to detect unfavorable developments early on. 
• A traffic-light system for mitigating risk from induced seismicity based on real-time data (using both classical and real-time adaptive approaches) will be developed and installed. A detailed risk mitigation protocol will take effect if seismicity exceeds conservative, pre-defined limits.
• The major risk in our experiment is to the people in the tunnel and lab. During the major fault stimulation experiments, the lab and tunnel will be subject to short but strong shaking because of its close proximity (within 50 – 100 meters) to the earthquake. The experiments will be remotely controlled from the outside of the tunnel to avoid risk to the people. 
• The seismic risk in this isolated mountain area is generally low. Even a magnitude 3 to 4 event, which would generate waveform amplitudes 100 – 1000 times larger than a magnitude 1, would not have significant consequences at the surface or at the Matterhorn Gotthard tunnel.