The storage of fluids in the subsurface is critical for a broad spectrum of applications including managed aquifer recharge, storage of liquified carbon dioxide and hydrogen, geothermal heat extraction and management of hydrocarbon resources. It is surprising then, that there has been relatively little measurement of the vertical distribution of fluid storage in geologic formations as compared with permeability. We present experiments in which fluid was injected into an important regional aquifer and the depth-dependent strain response measured using fiber optic distributed acoustic sensing. The formation expansion and contraction in response to fluid injection were measured in nanostrain. Strain, and the implied storage distribution, was highly localized in specific strata and demonstrated complex, (non-elastic and non-local) hydromechanical behavior. This new window into fluid-geomechanical coupling undermines some typically use models and observations, currently in practice, but provides potential for complete representation and prediction of fluid storage in the subsurface.

A significant increase in using distributed acoustic sensing (DAS) as a main sensor of choice in borehole seismic applications prompts more fundamental research in understanding what DAS measures, the limitations and performance of the technology, and how it compares to the conventional sensors. Here we use borehole seismic data collected with conventional three-component geophone and distributed acoustic sensing to quantify and compare seismic measurements at the GeoLab facility at Curtin University campus (Perth, Western Australia). This facility allows repetition of the experiment using different sensors in controlled, stable conditions; deployment fibre cables as borehole and surface arrays; and utilisation of different types of seismic sources. In this presentation we compare the two datasets acquired with the conventional Sercel SlimWave VSP tool and Silixa iDAS v2. Data collected with geophones correspond to particle velocity measurements and can be calibrated to this velocity from native system units (mV) to m/s. By differentiating the measurements over the 10-meter interval, we get “converted geophone” data, which now has a property of the strain rate with unit [1/s]. In addition to this, DAS data which natively measure phase variation over time, can be calibrated to the absolute strain rate with units [1/s]. Although calibrated to the same property, these two datasets are impacted by different factors that could affect their amplitudes’ absolute values. For example, the geophone amplitudes are affected by the type of geophone and its performance, probe’s housing, the quality of the probe’s coupling to the formation or the casing, etc. The DAS amplitudes are affected by cable design, the cable’s coupling to the formation, optical parameters, interrogator design, etc. We use both the peak-time amplitudes and the entire wavefield to compare the absolute values of the strain rate of both types of sensors. For the given study area and survey design (local geology, type of geophone and fibre-optic cable), it appears that amplitudes of the strain rate have similar absolute values.

Permanent reservoir surveillance is an invaluable monitoring tool for CO2 storage projects, as it tracks spatial-temporal evolution of the gas plume. The frequent images of CO2 plumes will facilitate history-matching of the reservoir simulations and increase confidence of early leakage detection. However, continuous data acquisition and real-time interpretation require a new approach to data analysis. Here we propose a data-driven approach to forecasting future time-lapse seismic images based on the observed past images and test this approach on the Otway Stage 2C data. The core component of the predictor is a convolutional neural network, which considers subsequent plume maps as colour layers, similarly to standard red-green-blue blending. Based on the spatial distribution of these ‘colours’ we may predict the future contour of the seismically visible part of the plume. The neural networks absorb the physics of CO2 migration through training on reservoir simulations for a wide range of injection scenarios and subsurface models. Extensive testing shows that realistic plumes for Stage 2C are too complicated and the neural network should be pre-trained on simpler reservoir simulations that include only one or two geological features, such as: faults, spill-points. Such staged training enables a gradual descent of the neural network optimization to a global minimum. In an upshot, the proposed algorithms are proven accurate. The approach is practical, because each CO2 storage project requires extensive pre-injection reservoir simulations. Once the predictor has been trained, it forecasts plume evolution almost instantly and quickly adapts to changing dynamics of the CO2 migration.