Skip to Main Content
  1. Home
  2. Search
  3. A Hybrid Analytical and Machine Learning Workflow for Completion Design Optimization in Unconventional Reservoirs
View
Download
Communities and Collections
Usage
  • 13 views
  • 28 downloads

A Hybrid Analytical and Machine Learning Workflow for Completion Design Optimization in Unconventional Reservoirs

  • Author / Creator
    Moussa, Tamer

  • Hydraulic fracturing combined with horizontal drilling is the key to unlocking tight reservoirs. However, understanding the relationship between reservoir characteristics, completion design and well productivity remains challenging. In the past decade, over 40,000 multi-fractured horizontal wells (MFHWs) have been completed in the Western Canadian Sedimentary Basin (WCSB). Despite completion intensity surging by nearly 100%, hydrocarbon productivity only rose by 20%. Why doesn't hydrocarbon production growth align with the increase in completion intensity? Is the efficiency gap attributed to formation characteristics, injection fluid properties, or the fracturing strategy?
    The main objective here is to develop a comprehensive analytical and machine learning (ML) workflow to evaluate and predict the recovery performance of MFHWs as a function of reservoir characteristics and completion design, and to optimize the completion design based on reservoir characteristics to maximize the productivity of MFHWs.
    While there have been advancements in the ML-based modelling approaches to predict well performance in tight reservoirs, there are limited studies focus on the comprehensive development of these resources considering both the reservoir characteristics and completion design. Therefore, this study seeks to address two main questions: i) how can the optimal sweet spots for MFHWs be efficiently identified? And ii) how can the completion design be optimized based on the reservoir quality and geomechanical properties? By addressing these questions, this thesis aims to provide a more holistic and effective approach for optimizing well performance in tight reservoirs.
    To achieve this objective, several new approaches are proposed to build this workflow. Initially, an iterative method is proposed to estimate dynamic fracture volume, porosity, and compressibility based on downhole pressure. Following this, a characteristic fracture closure rate is derived to describe the rate at which the effective fracture volume decreases during flowback. Subsequently, a water-oil-ratio model (WORM) is introduced to explain the observed log-linear relationship between WOR and load recovery as an analogy to the log-linear relationship between the water/oil relative-permeability ratio and water saturation. The coefficients from WORM are then coupled with key petrophysical properties using a neural network to predict WOR as a function of load recovery, forecast ultimate load recovery, and estimate effective fracture volume. Another data-driven model is proposed to predict oil production as a function of load recovery during the matrix-dominated flow regime. A support vector machine model is also developed to predict permeability from well log data, which facilitates the creation of high-resolution 3D maps of different petrophysical properties across the Montney formation, utilizing data from over 14,000 oil and gas wells. Similarly, sonic log data are utilized to estimate formation fracability, which is then interpolated using 3D kriging across the Montney formation. These developed petrophysical properties are incorporated to derive a Reservoir Quality Index (RQI), serving as a unified measure of reservoir quality based on petrophysical properties. Then, a series of ML-based proxy models are designed, trained, and proposed to correlate the oil and gas productivities of more than 10,000 oil and gas MFHWs with reservoir characteristics and completion design, alongside their mathematical representation for broader applicability. Finally, a supplementary study is proposed to explore the geothermal potential of suspended oil and gas MFHWs completed in WCSB. The aim is to identify the most suitable candidates for repurposing, especially considering the substantial investment initially made to complete these wells.
    The key results from this work show that the functional dependence of well productivity on completion-design varies depending on reservoir quality. In low-quality reservoirs, the effect of completion-design on productivity is less pronounced and the productivity follows reservoir quality. However, in high-quality reservoirs, the effect of completion-design becomes more significant, and the productivity can be reduced due to inefficient completion-design. Moreover, the productivity can be maximized by less intense completion-design in low-quality reservoirs. However, in high-quality reservoirs, intense completion significantly improves the productivity. Additionally, the application of a completion design to achieve a similar effective fracture volume on child wells does not necessarily lead to similar oil productivity compared to parent wells. It also depends on the quality of the reservoir at which the parent/child wells are completed. Finally, completion design parameters generally exhibit a greater influence compared to formation characteristics on both effective fracture volume and well productivity.

  • Subjects / Keywords
  • Graduation date
    Spring 2024
  • Type of Item
    Thesis
  • Degree
    Doctor of Philosophy
  • DOI
    https://doi.org/10.7939/r3-7ys6-pa67
  • License
    This thesis is made available by the University of Alberta Libraries with permission of the copyright owner solely for non-commercial purposes. This thesis, or any portion thereof, may not otherwise be copied or reproduced without the written consent of the copyright owner, except to the extent permitted by Canadian copyright law.