Discovery of Hidden Structures in Microseismic Data Using Tensor Decompositions and Multiway Component Analysis
Abstract
Microseismic data is used by companies to analyze and check numerous processes, including horizontal well performance, directional drilling and, most recently, hydraulic fracturing. The microseismic data analysis approach is important and there is a lot more to discover within microseismic technology with an application of the correct data analysis approach. For example, different visualization methods could potentially contain hidden structures, not visible by traditional methods. This work proposes a new methodology to access those hidden structures. In particular, machine learning tools, such as Tensor Decomposition (TD) and Multiway Component Analysis (MWCA), were utilized to gain more information from a previously existing pool of microseismic data. The extracted hidden structures can be used to learn more about source location, from which the information about fracture propagation could be inferred within the reservoir. This potentially gives a fast and cost-effective technique to analyze hydraulic fracturing processes. The work further illustrates applicability to a real microseismic study of the noise reduction and model reduction methods, based on the same machine learning techniques. A special case of TD, Higher-Order Singular Value Decomposition (HOSVD) is used to decompose the data, while MWCA is used to show the relationship between the decomposed structure and hidden structures within the dataset. Finally, possible steps to improve the technology are outlined. Since the applications of MWCA and TD are still emerging, future enhancements to this methodology are expected.
Citation
Yatsenko, Maxim Andreevich (2019). Discovery of Hidden Structures in Microseismic Data Using Tensor Decompositions and Multiway Component Analysis. Master's thesis, Texas A&M University. Available electronically from https : / /hdl .handle .net /1969 .1 /187911.