Fiber-optic cables have recently gained popularity for use as Distributed Acoustic Sensing (DAS) arrays for borehole microseismic monitoring due to their physical robustness as well as high spatial and temporal resolutions. As a result, the sensors record large amounts of data, making it very difficult to process in real-/semi-real-time using the conventional processing routines. We present a novel approach, based on deep learning, for handling the large amounts of DAS data in real-/semi-real-time. The proposed neural network was trained on synthetic microseismic data contaminated with real-ambient noise from field data and was validated using field DAS microseismic data obtained from a hydraulic fracturing operation. The results indicate that the trained network is capable of detecting and locating microseismic events from DAS data and simultaneously update the velocity model to a high degree of precision. The mean absolute errors in the event locations and the velocity model parameters are 2.04, 0.72, 2.76, 4.19 and 0.97 percent for distance (x), depth (z), P-wave velocity, S-wave velocity and density, respectively. In addition to automation and computational efficiency, deep learning reduces human expert data handling during processing, thus preserving data integrity leading to more accurate and reproducible results.

Microseismic monitoring is a useful enabler for reservoir characterization without which the information on the effects of reservoir operations such as hydraulic fracturing, enhanced oil recovery, carbon dioxide, or natural gas geological storage would be obscured. This research provides a new breakthrough in the tracking of the reservoir fracture network and characterization by detecting the microseismic events and locating their sources in real-time during reservoir operations. The monitoring was conducted using fiber optic distributed acoustic sensors (DAS) and the data were analyzed by deep learning. The use of DAS for microseismic monitoring is a game changer due to its excellent temporal and spatial resolution as well as cost-effectiveness. The deep learning approach is well-suited to dealing in real-time with the large amounts of data recorded by DAS equipment due to its computational speed. Two convolutional neural network based models were evaluated and the best one was used to detect and locate microseismic events from the DAS recorded field microseismic data from the FORGE project in Milford, United States. The results indicate the capability of deep neural networks to simultaneously detect and locate microseismic events from the raw DAS measurements. The results showed a small percentage error. In addition to the high spatial and temporal resolution, fiber optic cables are durable and can be installed permanently in the field and be used for decades. They are also resistant to high pressure, can withstand considerably high temperature, and therefore can be used even during field operations such as a flooding or hydraulic fracture stimulation. Deep neural networks are very robust; need minimum data pre-processing, can handle large volumes of data, and are able to perform multiple computations in a time- and cost-effective way. Once trained, the network can be easily adopted to new conditions through transfer learning.

Locating microseismic events is crucial to monitoring fracking activities, CO2 injection, and reservoirs in general. However, the process of locating such events is challenging, especially in low signal-to-noise scenarios. Thus, we introduce a conditional diffusion model for direct inversion of 3-D microseismic event locations from relatively irregularly spaced noisy waveform gathers, eliminating the need for manual arrival-time picking and reducing sensitivity to noise and velocity model uncertainty. The influence of the unknown event origin time on the waveform gathers is eliminated by cross-correlating every trace from the gather with a reference trace from the same gather. The 3-D microseismic event location coordinates (x, y, z) are projected onto two 2-D (X-Z and Y-Z) Gaussian heatmaps, enabling the use of a 2-D conditional diffusion model for the inversion task in contrast to traditional localization techniques. The model is trained to generate accurate microseismic event locations conditioned on microseismic waveform gathers. Tests on semi-synthetic data from the ToC2ME project demonstrate the model’s accuracy, robustness to noise, and computational efficiency. Furthermore, results from a real field microseismic event indicate high stability in the predicted event coordinates. These findings confirm the feasibility of using conditional diffusion models for reliable field-scale event localization.

Locations and source mechanisms of microseismic events are very crucial for understanding the fracturing behavior and evolution of stress fields within the reservoir and hence facilitates the detection of hydraulic fracture growth and estimation of the stimulated reservoir volume (SRV). In the classic workflow, there are two main methods for locating microseismic events with a calibrated fixed velocity model: grid search and linear inversion. The grid search is very stable; can find a global minimum and does not need initial event locations. However, it is computationally intensive and its resolution depends on the grid size, hence, it is not suitable for real-time monitoring. On the other hand, although the linear inversion method is quite fast, the inversion may be pushed into a local minimum by thin shale layers and large velocity contrasts leading to false locations. The source mechanisms of the located events, which provide information about the magnitudes, modes and orientations of the fractures, are obtained through moment tensor inversion of the recorded waveforms. In this paper, we propose a deep neural network approach to solve the above challenges, in real-time, and increase the efficiency and accuracy of location and moment tensor inversion of microseismic events, induced during hydraulic fracturing. Location of microseismic events was considered as a multi-dimensional and non-linear regression problem and a multi-layer two-dimensional (2D) convolutional neural network (CNN) was designed to perform the inversion. The source mechanisms of the microseismic events were inverted using a multi-head one-dimensional (1D) CNN. The neural networks were trained using synthetic microseismic events with low signal to noise ratio (SNR) to imitate field data. The overall results indicate that both the 2D CNN and 1D CNN models are capable of learning the relationship between the events locations and source mechanisms and the waveform data to a high degree of precision compared to classical methods. Both the event location and source mechanism errors are less than few percent. Deep learning offers a number of benefits for automated and real-time microseismic event location and moment tensor inversion, including least preprocessing, continuous improvement in performance as more training data is obtained, as well as low computational cost.