Quantifying Uncertainty for In-situ Stress Estimates using Reservoir Geomechanical Pressuremeter

• Author / Creator

  Zheng, DongMing

• Over the last two decades, several high-profile caprock performance issues, such as surface steam and fluid releases, have highlighted the importance of caprock integrity assessments and the critical role that robust determination of the initial in-situ stress tensor plays in these assessments. Traditionally, the minimum in-situ stress, which is a key component of the in-situ stress tensor, is determined from diagnostic fracture injection tests or minifrac tests. However, it is challenging to select the minimum stress from these tests. Consequently, recent research has pursued alternative techniques to help constrain the values of the in-situ stresses, resulting in the development of a reservoir geomechanical pressuremeter (RGP) that will allow for an integrated assessment of the in-situ reservoir rock compressibility and the direction and magnitude of maximum and minimum horizontal stresses. A conventional high-pressure pressuremeter was modified for deployment in a borehole using industry-standard wireline technology. RGP field tests were conducted at the Primrose Site project in 2016. Five intervals in three formations - Westgate, Joli Fou, and Clearwater - were tested with the deployment of the RGP tool. Interpretation and analysis of these data can provide vital information for the oil and gas industry, such as the shear modulus, undrained shear strength, and orientation and magnitude of anisotropic in-situ stresses.
  The frequentist and Bayesian statistical methods proposed were first applied to the uncertainty quantification of in-situ horizontal stress using a self-bored pressuremeter (SBP). The statistical methods used in the SBP test were then used to analyze the RGP tests. Using raw data from the Primrose-Wolf Lake project , uncertainties were first identified, followed by data conversion and
  corrections. Using the corrected RGP data, analytical and numerical models were coupled with optimization algorithms to find the best parameter estimates. To address the problem of non-unique solutions, uncertainty analyses were conducted using frequentist statistical assessment methods. With mean, standard deviations, and confidence intervals, uncertainties from parameter estimation were quantified, and non-uniqueness issues were addressed. Alternatively, Bayesian inference methods were adopted to evaluate in-situ horizontal stresses and material properties under a Bayesian statistical framework.
  To account for the radial and azimuthal anisotropies of borehole material, the modified strain-hardening/softening model was implemented in the statistical analysis for RGP tests in deep geological formations. The advantages of this model over the Mohr-Coulomb and conventional strain-hardening/softening models were verified through the interpretation of triaxial compression tests, demonstrating superior prediction accuracy, validity, and applicability.
  Compared with conventional pressuremeter interpretation methods, the proposed frequentist statistical inverse analysis methods can quantify the potential uncertainty and errors from ground properties and in-situ horizontal stress. In addition, the proposed Bayesian approach can continuously update one’s beliefs with new data through an open system, which is superior to the frequentist statistical methods employed in pressuremeter studies. The statistical methodology described in this study can be extended to other engineering inverse analysis problems, such as the calibration of constitutive models and inverse analysis of in-situ stress fields for horizontal drilling, tunnelling, and hydraulic fracturing.

• Subjects / Keywords

  • Pressuremeter
  • In-situ Stress
  • Uncertainty quantification
  • Frequentist statistical methods
  • Bayesian inference

• Graduation date

  Fall 2024

• Type of Item

  Thesis

• Degree

  Doctor of Philosophy

• DOI

  https://doi.org/10.7939/r3-vvan-qs61

• License

  This thesis is made available by the University of Alberta Library 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.