Humans can create earthquakes on natural faults by injecting fluids underground. However, details regarding what factors affect these earthquakes are not fully understood. We conducted laboratory experiments on 3 m blocks of rock that slip similar to a natural fault. Our experiments investigated how the initiation and overall size of earthquakes differed when fluid was pumped into a critically loaded fault (nearly ready to host an earthquake) versus a fault that was less critical. In the near-critical case, earthquakes occurred quickly and ruptured the entire fault. These earthquakes required fluid pressure to start the earthquake, but then were sustained by energy already present in the rock rather than due to fluid pressure. However, when the fault was not critical, earthquakes could only initiate after fluid pressure caused silent slip to redistribute significant amounts of shear stress from within the fluid pressurized area to the surrounding areas. When stress had redistributed enough that the surrounding areas reached a critical state, an earthquake initiated, but it did not rupture very far and remained small. Our experiments agree with recently published computer simulations that illustrate how induced earthquakes are strongly affected by the levels of preexisting stress in the rock.
