A Bayesian methodology for real-time updating of well trajectory in depositional space - Archive ouverte HAL Accéder directement au contenu
Communication Dans Un Congrès Année : 2022

A Bayesian methodology for real-time updating of well trajectory in depositional space

Résumé

Horizontal drilling is useful to target subsurface formations for injection or recovery of fluids in subsurface reservoirs. However, drilling with a low angle relatively to the stratigraphic dip can be a significant cause of targeting error and sub-optimal drilling process. Real-time consistency between the drill bit and the subsurface model is, therefore, essential for the drilling engineers. We propose a dynamic Bayesian updating of the positioning of a well relatively to a geological model, which accounts for: (1) A prior probabilistic model for petrophysical properties in the vicinity of the well being drilled. (2) The prior well geometric uncertainty based on deviation survey parameters. (3) Logging while drilling (LWD) measurements which can be compared to the prior geological model. Such models have recently been proposed in the literature, but they often consider a one-dimensional “typelog”, which is limited to represent lateral petrophysical variations. They also use relatively simplified geometric settings, making isopach layer assumptions. In this project, we consider the updating problem in depositional coordinates and a three-dimensional prior petrophysical model as derived from geostatistical reasoning. We demonstrate the proposed framework on a synthetic case study, both deterministically to find the most likely well trajectory or stochastically to generate possible trajectories, and discuss its potential towards implementing a model updating operator for geosteering. Results highlight that the process allows to rapidly update relative well trajectories honoring prior information and suggest that the method could be used to automatically detect some unseen geological objects such as faults not visible on seismic images.
Fichier non déposé

Dates et versions

hal-03985444 , version 1 (13-02-2023)

Identifiants

  • HAL Id : hal-03985444 , version 1

Citer

Julien Herrero, Guillaume Caumon, Paul Baville. A Bayesian methodology for real-time updating of well trajectory in depositional space. 21st Annual Conference of the IAMG, International Association for Mathematical Geosciences, Aug 2022, Nancy, France. ⟨hal-03985444⟩
19 Consultations
0 Téléchargements

Partager

Gmail Facebook X LinkedIn More