Four-dimensional seismic uncertainty quantification of the Hebron Field, Offshore Newfoundland, Canada

Abstract

The time-lapse or four-dimensional (4D) seismic method is used to image subsurface changes within a reservoir through acquiring multiple three-dimensional (3D) seismic surveys over the same area some time apart. The concept extends from the assumption that any changes in the seismic signal between surveys are due to perturbations in subsurface properties (i.e. fluid saturation, pressure, and temperature) caused by the production of a reservoir interval. In this thesis, we determine velocity estimates for a synthetic production-related time-lapse seismic anomaly in the Pool 1 reservoir of the Hebron Field located within the Jeanne d’Arc Basin, offshore Newfoundland, Canada. We use a time-lapse seismic uncertainty quantification algorithm to compute 1,000,000 velocity estimates of the time-lapse seismic change in the Pool 1 reservoir. We compute these models across four Markov chains with an acceptance rate of 13.5 - 16.0% in 5.4 days on a standard desktop (i7 6700 3.4 GHz processor and 16 GB memory). Repeated sampling generates end ranges of possible velocity models representing the synthetic production-related time-lapse change created for Pool 1 in this thesis. Quantities of interest (QoI) are defined to highlight the ability of these velocity models to represent the size and magnitude of the true synthetic production time-lapse change. We plot statistical versions of the velocity models recovered by an arbitrary Markov chain to visually showcase the recovered mean and maximum likelihood model representations of the true synthetic time-lapse velocity perturbation.