Empirical Analysis and Prognosis of the Induced Seismic Activity Rate of a Depleting Gas Reservoir in Time and Space

The Groningen gas field with a GIIP of 2900 BCM is one of the largest onshore gas-fields in the world. Production started in 1965 and to date (April 2016) some 74% has been produced. The first seismic event was registered in December 1991 and to date 266 events with magnitude 1.5 and above have been registered (1). Direct analysis of different parameters versus time did not result in a single dominant parameter that could be correlated to seismicity. Ref. (2) suggested to analyze the cumulative production volume versus the cumulative seismic events. This way time is eliminated as a variable and short time fluctuations are suppressed. The data can be well described by a quadratic equation with only two empirical parameters after introducing the volume produced at the first event, a parameter which follows directly from the production record and can be determined by geomechanical models (3). Analyzing the goodness of fit reveals a 95% confidence level of 8.7 events. The empirical model predicts that the event rate for various production volumes increasing with time. Alternatively, the inverted model shows that a constant activity rate requires a decreasing production rate. The time delay between the seismicity rate and the production rate was analyzed through the correlation coefficient between the two parameters. The correlation coefficients were relatively low and changing over time. Consistent maxima were found at 0 months, 3 to 5 months and 9 to 10 months. Estimations based on the characteristic diffusion time on the cluster scale (4) gives an increasing delay time with pressure depletion from 3.4 to 10 months. Nevertheless, the zero-delay analysis provides solid results giving confidence in quantification of the activity rate on the field scale. To be able to optimize the seismic risk related to regional production control the field is divided into several regions, based on the large offset faults and production experience. The two empirical parameters, the first-event volume and time delays are determined for each region separately. The confidence levels are much lower since the analysis is based on much sparser data. Nevertheless, we believe that this methodology shows the way forward in order to improve our understanding of the location-dependent seismic risk through production control in a depleting gas reservoir.