Reservoir Characterization of Plover Lake Heavy-Oil Field

Larry Lines¹, Joan Embleton¹, Mathew Fay¹, Steve Larter¹, Tony Settari¹, Bruce Palmiere², Carl Reine², and Douglas Schmitt³
¹CHORUS, University of Calgary, Calgary, AB, Canada - [email protected]
²Nexen Inc., Calgary, AB, Canada
³CHORUS, University of Alberta, Edmonton, AB, Canada

Short abstract

Enhanced production of heavy oil from the Cretaceous sands of Eastern Alberta and Western Saskatchewan presents many challenges – requiring a more complete description of lithology, porosity, permeability and changes in reservoir fluid composition and physical properties. Our reservoir projects near Plover Lake, Saskatchewan seek to produce reservoir models that are consistent with all available data including well logs, cores, produced fluids and seismic data. Thus far, we have effectively used dipole sonic data and multicomponent 3-D data to effectively delineate sand layers. Core measurements suggest that interbedded shale layers will impact vertical permeability and consequently oil production. In order to effectively map production and reservoir changes, we propose to use time-lapse (4-D) seismic surveys to update our reservoir models. These seismic measurements are coupled to laboratory measurements of Vp/Vs from core samples and detailed oil-column profiling of fluid properties. Experience with 4-D seismic data at nearby Bodo field, near Provost, Alberta, has shown that seismic monitoring can effectively map the reservoir changes due to cold production. Hence, we advocate a reservoir characterization strategy that involves the use of logs, cores and a base 3-D seismic survey to describe geology with repeated multicomponent 3-D surveys being used to map reservoir changes. Our study shows reservoir studies on models and real data from the Plover Lake area, along with planned future research.

Introduction

This paper examines a combined geological and geophysical reservoir analysis for a heavy oil field near Plover Lake, Saskatchewan, where Nexen Inc. has applied both hot and cold production methods. Oil sands of the Devonian-Missippian Bakken Formation are found in NE-SW trending shelf-sand tidal ridges that can be up to 30 m thick, 5 km wide, and 50 km long. Overlying Upper Bakken shales are preferentially preserved between sand ridges. The Bakken Formation is disconformably overlain by Lodgepole Formation carbonates (Mississippian) and/or clastics of the Lower Cretaceous Mannville Group. Since sandstones have larger S-wave velocities (and hence lower VP/VS ratios) than shales, VP/VS maps help from multicomponent seismic data help to identify thickening sand layers within the target zone. This analysis of an initial 3D-3C survey is described in a recent paper by Lines et al. (2005). In the future, we plan to examine changes within the reservoir due to cold production by the use of time-lapse seismology.

For this study, the 3D-3C seismic data was acquired by Veritas DGC using the VectorSeis® digital multicomponent recording system over a 8 square kilometer surface area. Multicomponent interpretation is made possible by a few dipole sonic logs, as well as many sonic and density logs.

The estimated Vp/Vs maps in this study are largely based on traveltime methods. However, a recent paper by Dumitrescu and Lines (2006) uses AVO analysis and simultaneous inversion to provide high-resolution images of the heavy oil formations. Finally, we examine cores from this area to provide a fine scale description of rock property variations in the field.

Methodology and Preliminary Results

The traveltime method for creating VP/VS maps from multicomponent data that is both robust and straight-forward. Flat events on vertical stacks are predominantly PP reflections, but on radial stacks are mostly due to PS conversions. Hence, interval traveltimes from a radial component stack contain information about S-wave velocities, and together with the corresponding traveltimes from the vertical component stack provide us with the necessary information to calculate VP/VS, the ratio of P-wave to S-wave velocities.

Figure 1 shows traveltime picks for the vertical and radial components on seismic lines in the Plover Lake field. The exploration targets are sand ridges within the Bakken Formation at a depth of about 800 m. For our traveltime picking, we initially used the Sparky Coal of the Mannville Group (~780 m) as a reference horizon above the Bakken and the Torquay Formation (Devonian carbonate at ~830 m) below the Bakken. Unfortunately, the Torquay is difficult to interpret on the seismic data, and it turns out that better traveltime picks can be obtained from a slightly deeper reflection (“base event” at ~1000 ms, PP time).The process of picking traveltimes was guided by the use of dipole sonic logs. These logs allowed us to compute PP and PS synthetic seismograms to aid interpretation of the Sparky and Torquay horizons throughout the vertical and radial 3D volumes.

The resulting VP/VS maps produced a very interesting and encouraging result for lithology discrimination. On the northern half of the map shown in Figure 2, we have marked enclosed features with dark lines to indicate an eroded Lodgepole formation. In the same figure, we have also marked a boundary along the southeastern side of the map which defines the erosional edge of both the Bakken sand ridge and overlying Lodgepole formation. Low VP/VS values in the middle of the map correspond to thicker Bakken and Lodgepole, while higher VP/VS on the southeastern side of the map correspond to a zone where the Bakken sand and Lodgepole Formation have both been eroded. In summary, when this VP/VS map is compared to previous interpretations based on well data and conventional (vertical component) data, the correlation of the map to other sources of lithology information is excellent. It should be mentioned that two other VP/VS maps based on different horizon picks are very similar to Figure 2 - suggesting that traveltime mapping of VP/VS is very robust and reliable. Another encouraging note is that this VP/VS map is very similar to those obtained by Dumitrescu and Lines (2006) using AVO analysis.

The complete reservoir characterization involves going beyond analysis of logs and seismic data. An examination of core from the Plover Lake field is being completed in order to more completely understand the reservoir rock properties and the permeability barriers due to shale layers. The inhomogeneities in the reservoir can be more completely understood by examining core such as that shown in Figure 3.

By examining these core samples, we realize the possibility of permeability barriers and the need for more sophisticated reservoir models and the need for enhanced seismic resolution. Additionally, we need to more completely understand the reservoir changes by using time-lapse seismology and rock physics measurements to link time-varying seismic properties to reservoir conditions. Further experiments are being planned.

Conclusions and Future Work

The computation of VP/VS maps from a 3-D multicomponent seismic survey has been very interesting and useful in delineating lithology changes. However, it would be interesting to also characterize reservoir changes due to cold production. Such reservoir changes have been modeled numerically but require further verification through physical modeling. Due to subtle nature of production effects, it is our opinion that cold production reservoir effects could best be detected by repeated time-lapse multicomponent surveys. The differencing of time-lapse surveys should eliminate lithology effects and emphasize effects due only to cold production. For this reason, a time-lapse multicomponent survey is being proposed to answer the reservoir monitoring questions for the Plover Lake field. The time-lapse seismic results and the well information can be used to update the reservoir models.

Acknowledgements

The authors thank the Consortium for Heavy Oil Research by University Scientists (CHORUS) for support of this project. We especially thank Nexen Inc., a CHORUS sponsor, for permission to show results from their Plover Lake data. Finally, we thank Sensor Geophysical for processing the seismic data Hampson-Russell for the use of their PROMC software.

References

Dumitrescu, C. and Lines, L., 2006, Vp/Vs ratio of a heavy oil field from Canada, paper submitted to the 2006 CSPG-CSEG convention.
Lines, L., Zou, Y., Zhang, A., Hall, K., Embleton, J., Palmiere, B., Reine, C., Bessette, P., Cary, P. and Secord, D., 2005, Vp/Vs characterization of a heavy-oil reservoir; The Leading Edge, 1134-1136.

Figure 1. Comparison of PP (left) and PS inlines (right) at a time scale corresponding to a VP/VS ratio of 2.33 (IPS = inches per second). The PP section times are between 200 and 1500 ms, and the PS section times are between 600 ms and 2770 ms.