Towards real-time microseismic processing: Efficient and robust methods for event detection and automated arrival time picking

Abstract

Hydraulic fracturing (HF) is a method to enhance the production of crude oil and natural gas trapped in impermeable geologic formations. HF technology has the capability to unleash reserves that were previously considered uneconomical. Microseismic (MS) monitoring involves the detection and localization of microseismic events, including induced seismicity generated by HF. MS methods are used to aid in HF design optimization. A typical MS survey may last months and generates terabytes of raw data, motivating the use of automatic triggering and location algorithms. To address these challenges, workflows are developed in this thesis based on novel methods for event-detection, event-validation and arrival-time picking. These workflows are applicable for a variety of acquisition geometries and signal-to-noise ratio, as well as real-time applications and post-acquisition processing modes. The Tony Creek Dual Microseismic Experiment (ToC2ME), a MS program acquired in west-central Alberta by the University of Calgary within the Duvernay unconventional play, was used as case study. A detailed study of signal quality and noise was undertaken, leading to substantial improvement in processing results. The noise analysis showed that approximately half of the stations are strongly influenced by cultural noise, making their use counterproductive in detection algorithms but generally suitable for automatic arrival-time picking of the data from the station. A new algorithm developed for event-detection, called energy stack, provides a similar detection rate to previous method, but has advantages of having significantly faster execution speed, fewer parameters and the ability to detect events that were previously undetected. A new method for arrival-time picking uses the kurtosis derivative, a high-order statistical measure, in a sliding window. This approach yields accurate and precise arrival time picks under diverse noise conditions. In summary, when compared to other methods that are currently used, the developed workflows are robust in different noise environments, advantageous for implementation since they use fewer parameters, are computationally faster and require less human interaction. All of these constitute desirable features for real-time microseismic monitoring systems.