Quantifying Resources for the Surmont Lease with 2D Mapping and Multivariate Statistics
- Weishan Ren (ConocoPhillips Canada) | Clayton V. Deutsch (U. of Alberta) | David L. Garner (Chevron Canada Resources) | Emmanuel A. Mus (Total E&P USA, Inc.) | Thomas J. Wheeler (ConocoPhillips Co) | Jean-Francois Richy (ConocoPhillips Canada)
- Document ID
- Society of Petroleum Engineers
- SPE Reservoir Evaluation & Engineering
- Publication Date
- April 2008
- Document Type
- Journal Paper
- 341 - 351
- 2008. Society of Petroleum Engineers
- 5.1 Reservoir Characterisation, 5.6.1 Open hole/cased hole log analysis, 5.2.1 Phase Behavior and PVT Measurements, 1.6 Drilling Operations, 4.3.4 Scale, 5.1.5 Geologic Modeling, 4.1.2 Separation and Treating, 5.3.9 Steam Assisted Gravity Drainage, 5.8.5 Oil Sand, Oil Shale, Bitumen, 4.1.5 Processing Equipment, 7.6.2 Data Integration
- 1 in the last 30 days
- 564 since 2007
- Show more detail
- View rights & permissions
|SPE Member Price:||USD 10.00|
|SPE Non-Member Price:||USD 30.00|
The McMurray formation consists of heterogeneous Cretaceous-bitumen-saturated sands. The reservoirs are thick and laterally extensive in the main fairways. Many commercial projects are in the early stages of development. Resources too deep to mine are considering steam assisted gravity drainage (SAGD) (Butler 1991). Detailed high-resolution 3D geostatistical modeling is useful for individual well-pair or pad flow simulation, but is neither practical nor necessary for resource assessment across large areas.
A methodology for resource assessment is developed from a geostatistical study on the Surmont lease. The uncertainty in more than 30 correlated variables is calculated on a dense 2D grid using all available information including wells, seismic, and geologic trends. The correlation structure between the variables is modeled under a multivariate Gaussian model. The local distributions of uncertainty have been checked with cross validation and with more than 100 new wells drilled during the last two drilling seasons. Resource uncertainty across the entire lease area and a number of arbitrary development areas is derived from the 2D maps of uncertainty. A combined P-field/LU simulation approach is used; the global uncertainty is consistent with the local uncertainty.
The McMurray formation contains a large oil-sands resource. A small portion of oil sands can be recovered by surface mining; most of the bitumen resource will be produced by advanced heavy-oil-recovery technology, such as the SAGD process. Accurate estimation of the in-situ resource range and associated risks is important for reservoir planning and development.
Detailed 3D models of heterogeneity are useful. They provide numerical models consistent with small-scale well data, measures of connectivity, and visualizations that appear realistic. The challenge of 3D models in the context of our problem is two-fold: the size of the models, and the requirement for realistic summaries of reservoir quality at each location. The study area is more than 500 km2, the thickness is on the order of 100m, there are more than 10 variables of interest, and we would need 100 or more realizations to represent uncertainty. More than 20 billion numbers would need to be routinely manipulated to understand Surmont at a relatively coarse discretization of 50×50×1 m.
The second challenge is more subtle. Reservoir management decisions depend on many factors (such as the thickness of good-quality reservoir, presence of top- or bottomwater, structure of the base reservoir, and geological variability). These factors are, for the most part, areal summaries of the reservoir. They can be reliably calculated from the well data; however, they are not as reliably estimated from 3D models. High-resolution geostatistical models do not reproduce all of the complex geological features and trends. This challenge is addressed by research.
In summary, the advantages of using 2D geostatistical modeling include good estimates of reservoir quality consistent with available well data, uncertainty at each location, and simple and fast modeling of variables required for decision making. There are several geostatistical techniques that can be used to integrate different data into a geological model including Gaussian-based Bayesian updating (Doyen et al. 1996), collocated cokriging, and full cokriging (Deutsch and Journel 1998; Goovaerts 1997). The Bayesian updating approach is used because of its reliability and simplicity in data integration (Deutsch and Zanon 2004).
Several reservoir parameters are important. The thickness of net pay or net-continuous-bitumen (NCB) thickness is related to the height of an anticipated steam chamber. The bulk oil weight (BOW) measures the fraction of the bitumen mass to the total rock mass. The porosity, ø net , and oil saturation, S o, over the NCB are related to the recoverable bitumen by the SAGD process. An important feature of many areas of the McMurray is the presence of top water and top gas that can provide a sink for the injected steam and adversely affect recovery. These upper units are referred to as thief zones (TZs) for the injected steam. Each project and company identifies different critical parameters. The typical project will involve predicting 20 to 30 variables at each 2D location. Only a few variables will be described in this review paper. Most of the data are derived from well logs and core data.
The available data variables are divided into two types: primary variables that we must predict, and secondary variables that are established from the geophysical interpretation or geological trend mapping. Secondary variables are used to constrain the prediction of primary variables away from the well data. The secondary variables are often structural variables. Three structural surfaces will be used in this paper: the bottom surface of the McMurray (BSM) formation, the top surface of the McMurray (TSM) formation, and the Wabiskaw-McMurray surface (WMS), which is a maximum-flooding surface above the McMurray formation. These structural data are usually quite reliable because of their lateral continuity, and they are derived from a variety of data sources (well and seismic data). These three variables and the calculated gross thickness of the McMurray (GTM) are treated as independent secondary variables for the 2D modeling. A schematic workflow is given in Fig. 1 to illustrate each step for local uncertainty assessment and for global resource uncertainty.
|File Size||3 MB||Number of Pages||11|
Alabert, F.G. 1987. ThePractice of Fast Conditional Simulations Through the LU Decomposition of theCovariance Matrix. Mathematical Geology 19 (5): 369-386. DOI:10.1007/BF00897191.
Butler, R.M. 1991. Thermal Recovery of Oil and Bitumen. EnglewoodCliffs, New Jersey: Prentice Hall.
Deutsch, C.V. and Journel, A.G. 1998. GSLIB: Geostatistical SoftwareLibrary and User's Guide. New York City: Oxford University Press.
Deutsch, C.V. and Zanon, S.D. 2004. Direct Prediction of ReservoirPerformance with Bayesian Updating Under a Multivariate Gaussian Model. Paper2004-047 presented at the Canadian International Petroleum Conference, Calgary,8-10 June.
Doyen, P.M., den Boer, L.D., and Pilley, W.R. 1996. Seismic Porosity Mapping in theEkofisk Field Using a New Form of Collocated Cokriging. Paper SPE 36498presented at the SPE Annual Technical Conference and Exhibition, Denver, 6-9October. DOI: 10.2118/36498-MS.
Goovaerts, P. 1997. Geostatistics for Natural Resources Evaluation.New York City: Oxford University Press.
McLennan, J. and Deutsch, C.V. 2003. Guide to SAGD ReservoirCharacterization Using Geostatistics. Edmonton, Alberta: Center forComputational Geostatistics, University of Alberta.
Ren, W., Leuangthong, O., and Deutsch, C.V. 2005. Global ResourceUncertainty Using a Spatial/Multivariate Decomposition Approach. Paperpresented at the Petroleum Society's Canadian International PetroleumConference, Calgary, 7-9 June.
Srivastava, R.M. 1992. Reservoir Characterization WithProbability Field Simulation. SPEFE 7 (4): 927-937.SPE-24753-MS. DOI: 10.2118/24753-MS.