Uncertainty Quantification and Ranking for Steam-Assisted Gravity Drainage (SAGD) Well-Pair Placement in the McMurray Oil Sands, Canada

Garner, David¹, T. J. Wheeler¹, Emmanuel Mus², Jean-François Richy² (1) ConocoPhillips Canada, Calgary, AB (2) Total, seconded to ConocoPhillips Canada, Calgary, Canada,

The McMurray formation is a highly heterogeneous Cretaceous clastic reservoir with thick, bitumen-saturated sands. For the Surmont lease, SAGD is being implemented to extract the bitumen. The project initially targets twenty horizontal well pairs. The main goals in this study were to aid and improve the complex decision making for vertical placement of horizontal wells, for optimizing expected performance, and for ranking the horizontal well pairs. A detailed workflow was developed. It consisted of integrating multi-scale data, build­ing geological models and post-processing realizations to account for uncertainties in steam chamber development.

Due to the nature of the reservoir, the SAGD process and this particular evaluation of post-processing models, each well pair can be studied as an individual reservoir within the surrounding drainage volume. To decide on vertical well placement, post-processing of the detailed geostatistical simulations provided uncertainty measures on the reservoir parame­ters as a function of elevation. Among the parameters computed and compared were effec­tive well length, net-to-gross, facies proportions, producible volumes, and a vertical dis­count factor on stranded volumes. This approach was a key element of the final elevation decision for the development wells. A comparison was made to the purely deterministic geo­logical approach.

The proposed wells were ranked by performance. A probabilistic model was developed to capture the reservoir parameter uncertainties and simulate performance of each well pair at a specific elevation. The ranking model used the well placements, extracted reservoir parameter distributions, SAGD operating conditions, employing Butler theory in a Monte Carlo framework. Vertical well placement was further optimized as a result of this perform­ance based criteria. These probabilistic outcomes and rankings were used to decide which subsets of planned well slots to drill first.