Date of Graduation
Statler College of Engineering and Mineral Resources
Petroleum and Natural Gas Engineering
Due to the anisotropy and heterogeneous nature of unconventional reservoirs like shale, a comprehensive parametric study to optimize hydraulic fracture treatment for such reservoirs is a tough challenge, especially when natural fractures are present. Most of the current frac simulators do not consider the anisotropy of rock elasticity in the shales. Besides, using the fracture simulation linked with reservoir simulation for the parametric study to understand the impact of multiple different design parameters on fracture propagation and production is time expensive and low efficient. The study proposes a workflow including a new orthorhombic (OB) rock algorithm to interpret geomechanical properties and in-situ stresses of naturally fractured shale. Additionally, it involves a supervised random sampling algorithm for supervised machine learning. The workflow is applied in a producing well in Marcellus Shale. The whole independent stiffness coefficients, elastic moduli, and in-situ stress profiles of the OB rock for the target formation are solved and interpreted. The interpreted mechanical properties and stresses with the current well design are used in a P3D fracture simulator to acquire a conductivity matrix. The obtained conductivity matrix is inserted into a reservoir model for estimating the production and history matching. A database is then built based on a supervised random sampling algorithm. The generated database is used to train and validate an artificial neural network (ANN) model. Lastly, the trained ANN can conduct parametric studies to analyze and study multiple fracture parameters impacting the production. First, the new OB model successfully predicts a more accurate Young's modulus and stress profiles than the old models for the current naturally fractured shale. Furthermore, the new algorithm of the OB rock predicts a peak stress barrier and obvious stress contrast within the target formation. The result of the OB model is consistent with the observation of the microseismic events. Second, using the proposed supervised random sampling algorithm, generated around 200 "critical" sample cases, which were selected to form a training database. A supervised machine learning algorithm is implemented to examine high-dimensional multivariable fractured designs more time efficiently. The current new method of OB rock illustrates the importance of considering natural fracture induced anisotropy in performing geomechanical interpretation, which leads to better evaluating the drilling, the completion, and the hydraulic fracturing designs. Training and validating ANN with a limited number of cases can investigate and optimize a complex, multivariable engineering design task such as hydraulic fracturing in a short time and highest efficiency.
Alhemdi, Aymen Ab Ali, "A workflow for unconventional reservoirs optimization using supervised machine learning in conjunction with orthorhombic elasticity modeling" (2022). Graduate Theses, Dissertations, and Problem Reports. 11377.