Search

Abstract: To help monitor the evolution of stored CO2, we have made precision seafloor gravity measurements at 30 seafloor stations above the Sleipner CO2 plume in the years 2002, 2005 and 2009. Each epoch of gravity data has an intra-survey repeatability of about 3 μGal (standard deviation), obtained using state-of-the-art instrumentation on top of pre-deployed seafloor benchmarks, with typically three visits on each location during a survey. We used three relative quartz-spring Scintrex CG-5 gravimeters in a unique offshore instrument package. Ocean tidal fluctuations and benchmark depths were determined using both pressure gauges on the gravity survey tool and stationary reference pressure gauges on the seafloor. We analyzed and accounted for multiple sources of changes in gravity to obtain an estimate of in situ CO2 density. First, the injected CO2, 5.88 million tonnes during this time period, displaces denser formation water, causing a negative gravity change above the plume. This is the signal of interest for this study. At the same time, hydrocarbon gas production and water influx into the deep, nearby gas reservoir cause an increase in gravity of higher amplitude and longer wavelength. Finally, by observing vertical depth changes of the seafloor benchmarks between surveys to mm precision, we quantified vertical benchmark movements caused by sediment scouring. Some of the benchmarks have experienced more than 10 cm vertical movement over the 7 year duration of the experiment, and erosional topography can be seen in a >10 m broad area around some of the benchmarks. The shifting sediment can also cause a change in the observed vertical gravity gradient. We inverted the gravity changes for simultaneous contributions from: (i) injected CO2 in the Utsira Formation, (ii) water flow into the Sleipner gas reservoir, and (iii) vertical benchmark movements. We estimate the part of the change in gravity caused by CO2 injection to be up to 12 μGal. If we assume a geometry of the plume as seen in 4D seismic data, the best match to the 30 stations requires an average CO2 density of 720±80 kg/ m3, neglecting dissolution of CO2 into the formation water. While the CO2 in the Utsira Fm. at Sleipner is supercritical, it is fairly close to the critical point; therefore only a slight increase in temperature could lower the density significantly. Density is also sensitive to impurities, which make up 1–2% of the injected material at Sleipner and reduce the density slightly. In the absence of down hole gauges in the injection well, we estimate the well-bottom CO2 temperature to be 48 °C and pressure to be hydrostatic (∼105 bar). These conditions give a calculated density of 485±10 kg/ m3 at the perforation. Density is expected to increase away from the well as CO2 cools down from contact with the cooler formation, up to a maximum of about 710 kg/ m3. The distribution of temperature and density within the plume is difficult to model exactly, but most of the CO2 is expected to cool down to initial reservoir temperature (∼35.5 °C at the perforation) except for a central high-temperature region where CO2 is still near the injection temperature. Because the undisturbed formation temperatures and the injection temperature are fairly well known, the 2002–2009 gravity change can be used to constrain the rate of dissolution of CO2 into the formation water. Dissolved CO2 is invisible in seismic data. The contribution from gravimetric data could therefore be highly valuable for monitoring this process, which is important for long-term predictions of the CO2 stored in the Utsira Fm. We give an upper bound on the dissolution rate of 1.8% per year. [Copyright &y& Elsevier]

This study probes the association between fluid injection in enhanced geothermal systems and certain kinds of seismicity that may result from hydraulic fracturing occurring at depth using unsupervised machine learning. In April and May 2019, a distributed acoustic sensing borehole array at the Frontier Observatory for Research in Geothermal Energy site near Milford, Utah recorded seismic data during hydraulic injection stimulation of a nearby well. Using an autoencoder, a type of deep neural network, we reduce the dimensionality of spectrograms of the detected signals to a lower-dimensional latent feature space with just nine dimensions. Next, Gaussian mixture model clustering is performed on this latent feature space, assigning each detected signal to one of 7 classes. For each signal class, we examine spatiotemporal distributions of the clustered results and find that total detections exhibit a bimodal distribution with respect to channel depth. The shallow mode occurs between 250 and 500 m, and the deep mode is centered around 750 m. In the temporal distribution, clustering results show the two best-clustered signal classes exhibit weak or no correlation with injection-related activities. More generally, we demonstrate the ability to discern not just when and where signals are detected, but also what kind, thus enabling rapid and targeted data exploration and providing constraints on source mechanisms. [ABSTRACT FROM AUTHOR]