Back to search

West Virginia University

An Artificial Neural Network Approach to Predicting Formation Stress in Multi-Stage Fractured Marcellus Shale Horizontal Wells Based on Drilling Operations Data

Abstract

dc:description.abstract

The distribution of the anisotropic minimum horizontal stress, both in horizontal and vertical directions, is necessary for effective hydraulic fracture treatment design in Marcellus Shale horizontal wells. Typically, the minimum horizontal stress can be estimated sonic logs. However, sonic log data is not commonly available for the horizontal Marcellus shale wells due to the complexity and cost. The objective of this research is to predict the anisotropic minimum horizontal stress by utilizing drilling parameters including depth, weight-on-bit (WOB), revolution per minute (RPM), standpipe pressure, torque, pump flow rate, and the rate of penetration (ROP). More specifically, artificial neural network (ANN) models will be developed to predict the anisotropic minimum horizontal stress for a horizontal Marcellus shale well from the drilling and well log data. Artificial neural networks are particularly useful to identify complex relationships to predict the properties of unconventional formations. The available data from a Marcellus Shale horizontal well was collected and filtered to prepare data sets for ANN training, testing, and validation purposes. Two networks, for the vertical and lateral sections of the well, were developed. The preliminary results indicated that inclusion of lithology, gamma-ray, and bulk density well log data as inputs can improve the predictability of the networks. Finally, the networks were used to predict the anisotropic minimum horizontal stress in a different Marcellus shale horizontal well with the available sonic log data. To evaluate the applicability of the ANN models, the predicted stresses by the networks were compared against those estimated from the sonic logs. The predictions by both networks (vertical and horizontal) were found to be in close agreement with those estimated from the sonic logs. The results of this research can be utilized as a predictive tool to help fill in the need for an accurate estimation of static geomechanical properties including the minimum horizontal stress in Marcellus Shale horizontal wells and to improve the fracturing treatment design.

Degree

thesis:*
Name thesis:degree_name
MS
Level thesis:degree_level
Thesis
Discipline thesis:degree_discipline
Petroleum and Natural Gas Engineering

Author and committee

dc:creator, dc:contributor.*
Author dc:creator
  • Alawadh, Moudhi
Contributors dc:contributor
  • Mohamed El Sgher
  • Samuel Ameri
  • Kashy Aminian

Subjects

dc:subject × 9

Identifiers

dc:identifier.*

Chain of custody

source
Harvested from
West Virginia University
Base URL
researchrepository.wvu.edu/do/oai/
Last updated
2026-11-19
Source record
OAI-PMH GetRecord
citation

Alawadh, Moudhi. An Artificial Neural Network Approach to Predicting Formation Stress in Multi-Stage Fractured Marcellus Shale Horizontal Wells Based on Drilling Operations Data. Thesis thesis, https://doi.org/10.33915/etd.12193