Micro-seismic and leveling monitoring of a solution mining cavern collapse

To improve our understanding in large scale ground failure phenomenon induced by old underground mining works, a field experiment was undertaken by the partners of GISOS1in collaboration with the SOLVAY mining company: a solution mine was instrumented in 2004 previously to its collapse which was triggered in February 2009, as part of the mining scheme. This solution mine is located in the Lorraine salt basin, at Cerville-Buissoncourt southwards from Nancy, France. To monitor the cavern collapse, a permanent monitoring system was set up. This comprises a SYTGEM multi-parameter system featuring high resolution microseismic (5 1D geophones, 4 3D geophones, 40-1000 Hz) linked to ground surface monitoring (tacheometer and GPS high resolution measurements). The important amount of data transmitted for on-line processing offered daily insight of the evolution of the geological system. The early signs of unstable evolution were detected by high resolution microseismic monitoring during spring 2008 : a shift in microseismic background regime as well as recurrent microseismic episodes were associated to a general process of rock failure due to the cavern growing upwards. This was accompanied by a few episodes of massive roof falls, with no ground surface movement detected. During a second and last stage of evolution that lasted three days, the cavern collapse was triggered by brine pumping. Three main microseismic regimes were then observed, each being well correlated with changes in both the surface subsidence rate and the brine level in the cavern. The first peak of activity, due again to massive roof falls below the Beaumont dolomite bed, also initiated the brittle rupture of this bed. The second and major peak of activity marked its rapid and energetic failure; it also corresponded to an acceleration in the subsidence rate. The last peak of activity, a few hours later, weaker in terms of number of events and energy released, signed the final collapse. The complete dataset confirms the major role of the Beaumont dolomite bed in the site stability. Besides, if the preliminary analysis of the microseismicity patterns right before the general collapse, confirms that the acceleration in the microseismic activity follows a power law, it also shows that the b-value is a complex marker in such context. Even though local falls of b-value well correlate, as expected, to local increases in microseismic energy release, it seems inappropriate to use it for forecasting purposes: over a short period of time, several b-value decreases are observed while a continuous pumping regime is applied to the system. So, apparent b-value variations don't show clear precursory pattern, as observed in other case-study. Nevertheless, further investigations over the different stages of evolution of the cavern, including the collapse period, are needed to confirm this preliminary result