Date of Graduation
Statler College of Engineering and Mineral Resources
Petroleum and Natural Gas Engineering
Operating companies in the unconventional Marcellus shale play have all faced a similar and problematic issue, while attempting to produce natural gas over the last decade. Companies have quickly realized that not every perforation along their horizontal wells are producing gas. In fact, producing perforations are only ranging from 15%-70% of the total perforations along the horizontal wellbore . This unexplained issue results in millions of dollars in lost revenue per well, in addition to the sunk cost of paying for completions that are not actually yielding any produced gas.
What is causing these perforations to have no produced gas? There are many theories being researched in the private sector and academia including: stress shadowing, proppant type and concentration, sand-outs, unconventional reservoir modeling, and improved geosteering. While any and all of those situations may have an impact on production, this study will focus on one potential issue with shale wells that may be the root cause of this phenomenon: the anisotropic nature of shale. By nature, shale is highly anisotropic, which means that the physical properties of shale change significantly from point to point in the x, y, and z directions. This is caused by the laminar structure of the shale due to the shales formation, effecting properties in the z-direction, as well as widespread natural fracturing effecting properties in the x-y directions . Is it possible that the random and highly variable physical properties of the Marcellus shale are responsible for poor fracture propagation and production at various perforated clusters along the horizontal wellbore?
If the physical properties of the shale change considerably along the wellbore, then an area with favorable geomechanical properties for hydraulic fracturing such as Young’s Modulus and Poisson’s Ratio could quickly become unfavorable conditions simply a few feet away. An Artificial Intelligence and Machine Learning method called Fuzzy Logic C-Means Clustering can be used to identify these random changes in shale properties along the wellbore. This is done by gathering raw measured data from sources such as a sonic log or natural fracture log and allowing the AI algorithm to classify each half-foot of shales along the wellbore into groups of ‘like’ shales. These newly defined classifications of shales are grouped together to include shales with similar physical properties to each other. This can be used to identify areas of anisotropy along the wellbore that would have previously been unseen, allowing for an engineered completions design that ensures all perforated clusters will be placed against shales with similar physical properties. This is likely to result in improved overall production within a stage, since fractures would not be induced at different types or qualities of shales within a single stage. The theory is that a stage where every cluster successfully propagates a fracture will have higher production than a stage with one or two dominant fractures. The individual fractures may be smaller using this method, but the improved cluster efficiency could see improved production.
The use of C-Means Fuzzy Clustering is validated when clustering sonic and natural fracture log data for the MSEEL well MIP-3H, and comparing the changes in classification with a production log for the well. The changes in classification are quantified as an anisotropy indicator value (AIV). When comparing the AIV with the production, a peak and valley relationship is observed. When the AIV is high, the production is low or near zero at that given cluster. When the AIV is low, the opposite is true. In fact, the Fuzzy C-Means Clustering model was able to identify a high AIV at 88% of the non-producing clusters for MIP-3H. This suggests a strong correlation between the anisotropy of shale, and its effect on achieving a successful completions design.
The Fuzzy C-Means model can then be applied to a full horizontal wellbore sonic and natural fracture log in order to optimize a more successful completions design that is likely to see improved cluster efficiency when accounting for the shale anisotropy.
Palmer, Cole E., "Using AI and Machine Learning to Indicate Shale Anisotropy and Assist in Completions Design" (2020). Graduate Theses, Dissertations, and Problem Reports. 7870.