Seismic Source and Elastic Full-Waveform Inversion Using Distributed Acoustic Sensing and Perforation Shots in Unconventional Reservoirs SEP-193 (2023)
Table of contents
- Chapter 1: Introduction
- Chapter 2: Seismic modeling using SBP operators
- Chapter 3: Waveform inversion toolbox
- Chapter 4: Moment tensor of perforation shots
- Chapter 5: Reservoir properties estimation by extended FWI
- Chapter 6: 3D application using cross-well DAS data
- Chapter 7: Summary and conclusions
- Appendix A: Elastic wave analytical solutions
- Appendix B: Green's identities and FWI gradients
- Appendix C: Elementary moment tensors
- Appendix D: Least-squares matching filters
- Appendix E: Attenuation effects and Q estimation
Unconventional oil and gas resources are poised to maintain a significant presence in the global energy landscape for the foreseeable future. Nevertheless, our comprehension of unconventional reservoirs remains limited and has not fully harnessed advancements in seismic imaging technologies. This limitation arises from the substantial development costs associated with unconventional reservoirs and the misleading simplicity of their geological structures, which often lead to overly simplistic assumptions regarding their uniformity. The introduction of distributed acoustic sensing (DAS) technology has paved the way for various geophysical applications in these reservoirs through fiber optic deployment in horizontal wells. These applications primarily focus on characterizing fractures induced by hydraulic stimulation but fall short of providing a holistic characterization of the entire reservoir.
In this context, I propose harnessing the cost-effective yet invaluable DAS-recorded seismic waves generated by perforation shots within an elastic full-waveform inversion (FWI) framework. This approach aims to estimate high-resolution elastic properties of the reservoir formation, which can subsequently serve multiple purposes, including mapping heterogeneities, assessing potential hazards, refining reservoir models for resource evaluation, and monitoring stimulation and production operations.
This dissertation addresses two principal challenges in pursuit of this goal. Firstly, the representation of perforation shots as seismic sources, a prerequisite for FWI, remains undefined. I develop a source model based on a superposition of three mechanisms and derive the corresponding moment tensor representation. A workflow for estimating source parameters from field DAS data is established, with the moment tensor serving as an indicator of a perforation shot's effectiveness in creating micro-cracks in the surrounding rock. Secondly, the presence of seismic sources within the layer of interest introduces significant artifacts and footprints in the FWI results, particularly susceptible to source-related errors. To mitigate these issues, I propose a simple and effective method for footprint removal while robustly updating the model in their vicinity. This method leverages illumination redundancy and extends the model along the sources. Each source updates one component of the extended model, with a regularization term ensuring mutual consistency among these components, except within their respective footprints. After demonstrating the method's resilience in the presence of source-related errors, I apply it to a field DAS dataset, successfully identifying meter-scale reservoir anomalies indicative of lower pore pressure or denser shale regions. I further adapt this method to assess the long-term stimulation effects, revealing only a mild decrease in shear wave speed.
Finally, I extend my investigations to encompass source and elastic parameter estimation in a three-dimensional geometry using cross-well DAS/perforation data. The results highlight the prevalence of explosion and tensile crack mechanisms as dominant source mechanisms and emphasize significant lateral variations in elastic parameters, further underscoring the lateral heterogeneity of unconventional reservoirs.
Reproducibility and source codes
This thesis has been tested for reproducibility. The source code and reproducibility steps are available at the GitHub repository: https://github.com/nmbader/phd_thesis