Accepted Manuscript Lithofacies modeling by multipoint statistics and economic evaluation by NPV volume for the early Cretaceous Wabiskaw Member in Athabasca oilsands area, Canada Kwang Hyun Kim, Kyungbook Lee, Hyun Suk Lee, Chul Woo Rhee, Hyun Don Shin PII:
S1674-9871(17)30058-0
DOI:
10.1016/j.gsf.2017.04.005
Reference:
GSF 556
To appear in:
Geoscience Frontiers
Received Date: 5 October 2016 Revised Date:
1 April 2017
Accepted Date: 7 April 2017
Please cite this article as: Kim, K.H., Lee, K., Lee, H.S., Rhee, C.W., Shin, H.D., Lithofacies modeling by multipoint statistics and economic evaluation by NPV volume for the early Cretaceous Wabiskaw Member in Athabasca oilsands area, Canada, Geoscience Frontiers (2017), doi: 10.1016/ j.gsf.2017.04.005. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT Lithofacies modeling by multipoint statistics and economic evaluation by NPV volume for the early Cretaceous Wabiskaw Member in Athabasca oilsands area, Canada
RI PT
Kwang Hyun Kima,b, Kyungbook Leea, Hyun Suk Leea,*, Chul Woo Rheeb, Hyun Don Shinc
a
Petroleum and Marine Research Division, Korea Institute of Geoscience and Mineral Resources, Daejeon 34132, Republic of Korea
b
c
SC
Department of Earth and Environmental Sciences, Chungbuk National University, Cheongju 28644, Republic of Korea Department of Energy Resources Engineering, Inha University, Incheon 22212, Republic of Korea
AC C
EP
TE D
M AN U
* Corresponding author. E-mail address:
[email protected], Tel: +82-10-5168-7771
1
ACCEPTED MANUSCRIPT Abstract The static modeling and dynamic simulation are essential and critical processes in petroleum exploration and development. In this study, lithofacies models for Wabiskaw
RI PT
Member in Athabasca, Canada are generated by both multipoint statistics (MPS) with training image (TI) and then compared with those built by sequential indicator simulation (SIS). Three TIs are selected from modern depositional environments; the Orinoco River Delta estuary, Cobequid bay-Salmon River estuary, and Danube River delta environment. In order
SC
to validate lithofacies models, averages and variances of similarity in lithofacies are
M AN U
calculated through random and zonal blind-well tests. In random six-blind-well test similarity average of MPS models is higher than that of SIS model. The Salmon MPS model the mostly resembles facies pattern of Wabiskaw Member in subsurface. Zonal blind-well tests show that successful lithofacies modeling for transitional depositional setting requires additional or
TE D
proper zonation information on horizontal variation, vertical proportion, and secondary data. As Wabiskaw Member is frontier oilsands lease, it is impossible to evaluate the economics from production data or dynamic simulation. In this study, as a dynamic steam assisted
EP
gravity drainage (SAGD) performance indicator (SPIDER) on the basis of reservoir characteristics is calculated to build 3D reservoir model for the evaluation of the SAGD
AC C
feasibility in Wabiskaw Member. SPIDER depends on reservoir properties, economic limit of steam-oil ratio, and bitumen price. Reservoir properties like porosity, permeability, and water saturation are measured from 13 cores and calculated from 201 well-logs. Three dimensional volumes of reservoir properties are constructed mostly based on relationships among properties. Finally, NPV volume can be built by equation relating NPV and SPIDER. The economic area over criterion of US$10,000 is identified, and the ranges of reservoir properties are estimated. NPV-volume-generation workflow from reservoir parameter to static model provides cost and time effective method to evaluate the oilsands SAGD project. 2
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
Key words: Oilsands; Multipoint statistics (MPS); Reservoir static modeling; Economic evaluation; Net present value (NPV), SAGD, and SPIDER
3
ACCEPTED MANUSCRIPT
1. Introduction In petroleum exploration and production projects, a recovery factor and revenue are critical
RI PT
because any drilling program is expensive. In order to make a sound decision on the development plan, reservoir static models are built and utilized for dynamic simulations to find the best well location and to reduce a risk. These static models contain reservoir
SC
properties like porosity, permeability and saturation which are basic input data for dynamic simulation and are mainly controlled by lithofacies distribution. Lithofacies modelling has
M AN U
commonly been carried out by sequential indicator simulation (SIS) (Deutsch and Journel, 1992; Yao, 2002; Deutsch, 2006) that has some limitations on realistic description of facies connections, especially for those of complex reservoirs (Remy et al., 2009). Object-based Modeling method (OBM) can resolve aforementioned limitation of SIS. For instance, it is
TE D
appropriate method to simulate model of channel complex (Deutsch and Tran, 2002). However, OBM should need pretreatment process and spatial information of depositional element (Pyrcz and Deutsch, 2014). So, the application is considerably limited.
EP
Multipoint statistics (MPS) was developed by Guardiano and Srivastava (1993) to
AC C
overcome the limitation of SIS and MPS and then Strebelle (2000) suggested the procedure of obtaining probability from training image (TI) using search tree. Recently, MPS has been widely applied for reservoir modeling due to several advantages such as realistic facies distribution, honoring well data, easy conditioning with secondary data (Hashemi et al., 2014; Pyrcz and Deutsch, 2014; de Carbalho et al., 2016) and alternative of variogram modeling (Strebelle, 2002; Caers and Zhang, 2004). The variogram is one of the most important parameters in two-point algorithms. However, it is challenging to build variogram models for each facies, direction, and reservoir properties if a few well data is available. As such, this
4
ACCEPTED MANUSCRIPT study adopts TI of MPS rather than variogram in order to better represent spatial information. TI generally embodies geological concept for shape and distribution of lithofacies (Strebelle, 2000, 2002). TI is scanned through search window to reproduce a probability tree which is
RI PT
then utilized to build cumulative distribution function for simulation of each cell (Strebelle, 2002). Because, the selection of proper TI controls the result of facies modeling, validation of TI is essential for reliable property modeling.
SC
After reservoir property model is validated by facies model, it is utilized for dynamic simulation to establish a development plan such as well location and spacing. For example,
M AN U
steam assisted gravity drainage (SAGD), one of the widely used thermal recovery methods in oilsands reservoirs, uses well pair of steam-injection and oil-production wells. The position of well pair is a critical for oil recovery and can be determined by dynamic simulation. However, the simulation requires various petro-physical properties such as relative permeability, fluid
TE D
composition, and capillary pressure for reliable prediction. Also, development plan usually requires a huge number of dynamic simulations for sensitivity analysis and optimization scheme. Thus, it can be a huge burden for thermal forward simulation.
EP
To mitigate the simulation time, several researches have reducing the number of
AC C
simulation using ranking or clustering methods (Lee et al., 2013; Lim et al., 2014). Also, there are studies on a relationship between static parameters and dynamic responses using proxy model or experimental relationship. Shin (2008) proposed economic indicator for SAGD project in a reservoir scale as function of several reservoir parameters such as vertical thickness, porosity, vertical permeability, and oil saturation. In this research, the function, dynamic steam assisted gravity drainage (SAGD) performance indicator (SPIDER), is applied to unit cell to identify SAGD sweet spot directly from the property model without reservoir simulation. Therefore, property models based on reliable lithofacies modeling are
5
ACCEPTED MANUSCRIPT critical in locating the sweetspot for SAGD project. The Cretaceous Wabiskaw Member in Western Canadian Sedimentary Basin (WCSB) filled with bitumen but still remains a frontier oilsands formation due to thin reservoir
RI PT
thickness. The accurate assessment to this member via MPS and SPIDER will provide helpful information to increase economics of SAGD for frontier lease. As input data for geostatistics, hard data are prepared from well logging and TI candidates are chosen from geological
SC
setting and structural framework (Keith et al., 1988; Wynne et al., 1994; Wightman et al., 1997; Hein and Cotterill, 2006; Shields and Strobl, 2010). Then, the TIs are evaluated
M AN U
through several blind-well tests. Lithofacies models from MPS are compared with result from SIS. Property model built from the final lithofacies model is utilized for calculation of net present value (NPV) of each cell to obtain sweetspot model. Finally, the most proper MPS lithofacies model is validated and economic area in the frontier oilsands lease can be
2. Methodology 2.1. Workflow
TE D
suggested only based on geological characteristics of reservoir.
EP
Workflow for lithofacies modeling is designed to build model via either MPS or SIS and to
AC C
validate model by blind-well tests. Once available data are reviewed, depositional environment and structural framework are interpreted from literature review (Keith et al., 1988; Wynne et al., 1994; Wightman et al., 1997; Hein and Cotterill, 2006; Shields and Strobl, 2010) and well data. In this study, there are 201 well-logs and 13 core descriptions available. Spatial relationships, TI, and variogram are assessed for the two geostatistical methods. After the best lithofacies models are confirmed through random six-blind-well tests and zonal blind-well tests, the reservoir properties set for porosity, permeability, and oil saturation are simulated by Sequential Gaussian Simulation (SGS) and regression equation. To calculate the
6
ACCEPTED MANUSCRIPT monetary value of each cell in the Wabiskaw model, NPV volume is generated from the static models by SPIDER. Finally, sweetspot in the study area and range in reservoir parameters can be determined from NPV volume.
RI PT
2.2. Geological setting The Early Cretaceous Manville Group in the Athabasca oilsands region (Fig. 1) mainly consists of the McMurray, Clearwater, and Grand Rapid formations. The Wabiskaw Member
SC
is the basal member of the Clearwater Formation and unconformably overlies the McMurray Formation (Carrigy, 1967; Flach, 1984; Keith et al., 1988; Wightman et al., 1997; Hein and
M AN U
Cotterll, 2006; Benyon et al., 2014). The Member is correlated to the Bluesky Formation in the Peace River area (Shields and Strobl, 2010) and the Glauconitic Sandstone Formation in the Lloydminster area (Central Alberta) (Flach, 1984). The Member has been subdivided into B, C, and D beds based on erosional surface (Wynne et al., 1994; Wightman et al., 1997;
TE D
Alberta Energy and Utilities Board, 2003; Hein and Cotterill, 2006). The WCSB is the largest sedimentary basin in Canada. During the Aptian to Albian, transgression occurred from the Western Interior Sea (Boreal Sea) which is located on the
EP
northern Alberta State. During the Wabiskaw Member was deposited in the WCSB (Mossop,
AC C
1980; Wynne et al., 1994; Cant, 1996; Hein and Cotterill, 2006). The Wabiskaw Member was previously interpreted as the nearshore environment (as a bar deposits) (Carrigy, 1967) and recently revealed as the valley-filled deposits in low relief area (Wynne et al., 1994; Wightman et al., 1997; Hein and Cotterill, 2006). The lower bed of the Wabiskaw D beds was interpreted as a valley-filled deposits and shale beds which were interpreted as the shallow marine deposits (Wynne et al., 1994; Hein and Cotterill, 2006). Shields and Strobl (2010) recently interpreted that the Wabiskaw D beds formed in outer estuary to shallow marine environment during transgression. The study on Kirby North lease, 7
ACCEPTED MANUSCRIPT 250 km north from their study area suggests three facies associations in the Wabiskaw D beds: tidally influenced channel fill (FA1), high energy bay fill (FA2), and marine shoreface sediments (FA3). FA1 and FA2 of the Wabiskaw D Bed are considered to be good reservoir
RI PT
with high porosity, permeability, and high oil saturation (Table 1). In study area, located in the township 76, ranges 6–7 west of the 4th, the Wabiskaw D beds mainly occur in study area only (Jo and Ha, 2013; Shinn et al., 2014), whereas the D sand bed is not deposited in the
SC
township 76 and range 6 (Jo and Ha, 2013). Based on these facts, TI for the Wabiskaw D beds can be found from modern depositional environments of tide-dominated outer estuary to
M AN U
shallow marine with gradual changes from sand to shale deposits. 2.3. Lithofacies classification and structural framework
To prepare hard data for the geostatistical methods, all data from 201 wells are utilized. Six lithofacies are identified from the core description of 13 wells in study area, which can be grouped into four lithofacies associations (sand, shaly sand, sandy shale, and shale). Neural
TE D
network analysis applies to establish facies in boreholes without facies description on lithofacies or sedimentary facies (Wong et al., 1995; Negi et al., 2006; Qi and Carr, 2006;
EP
Dubois et al., 2007; Wang and Carr, 2012). These facies are then linked to four electro-facies using unsupervised neural network analysis with gamma-ray log in the same boreholes. The
AC C
result is shown of 87.9% similarity between lithofacies and electro-facies. Finally, 188 noncored well-logging data are converted to facies data using unsupervised neural network analysis.
Resistivity and gamma-ray logs are used with their relevant lithofacies to construct structural framework of the Wabiskaw Member. Well-logs are interpreted from 201 wells, whereas the lithofacies are derived from 13 core descriptions. The upper and lower boundaries of the Wabiskaw Member in study area are carefully traced based on the markers of gamma-ray and resistivity logs. These boundaries are confirmed by the convergent 8
ACCEPTED MANUSCRIPT interpolation method with well top picking. The net thickness of the Member decreases eastward from 35 to 7 m. Paleotopography is reconstructed through flattening of the upper boundary which represents thick shale deposition at maximum flooding time. It means that
RI PT
we could confirm locations of accommodation space when the Wabiskaw Member was formed. The Wabiskaw Member is deeper in the western area than in the eastern part. So, in the early transgressive stage, the sediment of Member would have been deposited in the
SC
western area. And then, accommodation area has been extended with sea level rising to east
3. Lithofacies modeling and validation 3.1. Selection of TI
M AN U
in study area.
Generally TI for specific stratigraphic unit (e.g. formation and member) can be obtained from either inverted or attributed seismic data, reconstructed depositional model or
TE D
corresponding modern environments (Barson et al., 2001; Maharaja, 2008; Pyrcz et al., 2008; Jung et al., 2012; Hashemi et al., 2014). In this study, images from seismic data are not available because the member is thinner than resolution of seismic data. So in this study, TI is
EP
designed based on specific depositional environments. The Wabiskaw Member was subdivided by the transgressive or erosive surfaces (Hein and Cotterill, 2006) and the lowest
AC C
Wabiskaw D sand bed is simulated as reservoir in study area. It is overlain by transgressive erosion surfaces and subsequent overlying regional marine shale (Hein and Cotterill, 2006). The Wabiskaw D beds are interpreted as formed in estuary sand to shallow marine shale during sea level rising (Shields and Strobl, 2010). Depositional environments of the Orinoco River Delta estuary (Orinoco) and Cobequid bay-Salmon River estuary (Salmon) can be suitable TIs due to similar with these sedimentological literature studies of the Wabiskaw D beds. The typical wave-dominated Danube River Delta (Danube) is also selected for TI as a control group. The TIs for depositional environment of the Wabiskaw Member are captured 9
ACCEPTED MANUSCRIPT on the satellite images on the Google Earth. These are 2D data without vertical depositional trend information. This trend is enough to realize with horizontal aspect because the Wabiskaw Member has considerably thin thickness and was deposited during transgressive
RI PT
environment only. Orinoco has been divided as central/northwest and southern sectors (Warne et al., 2002). The central/northwest sector mainly represents in the Orinoco strongly affected by littoral
SC
current, swamp and marsh systems near the coast. On the contrast, the coast of southern sector represents estuary environment since it was influenced by river and tide processes (Fig.
M AN U
3a) (Warne et al., 2002). Therefore, southern sector of the Orinoco is proper for TI (Fig. 3b). Sandy sediments were deposited near coast owing to decreased flow velocity, forming sand bar. Suspended muddy sediments were deposited in the proximal region from the source, because flow velocity was rapidly decreased by combined river and littoral current (Warne et
TE D
al., 2002). Red box on Fig. 3b is target sub-environment containing sand bar deposits and suspended mud (Fig. 3c). TI size is similar to actual size of oilsands lease (Fig. 2) and TI is designed to Fig. 3d.
EP
The Salmon represents typical macro-tide estuary (Fig. 4a) comprises three facies zones;
AC C
facies zones 1, 2 and 3 indicate lower, middle, and upper part of estuary environment (Fig. 4a) (Dalrymple et al., 1991). TI candidate in the Salmon contains the elongated sand bar deposits and shallow marine deposits (Fig. 4c). TI from Orinoco is shown for these sand bars and fine grain sediment depositional element (Fig. 4d). Both the Orinoco and Salmon TIs have sand bars developed in outer estuary environment and fine grain sediment deposits in shallow marine environment. Two TIs differ in the location of transitional zone. The Orinoco TI has distinctive shallow marine deposits, while the Salmon TI has predominant sand bars in outer estuary environment.
10
ACCEPTED MANUSCRIPT As the control groups, the Danube represents wave dominant delta environment (Fig. 5a) (Panin and Jipa, 2002; Giosan et al., 2005). TI from Danube consists of the river mouth sand and transitional zone between river mouth and shallow marine (Fig. 5b and c). Result of
RI PT
lithofacies modeling based on Danube TI is compared with those of the Orinoco and the Salmon. 3.2. Results
SC
Setting for MPS lithofacies modeling is shown in Table 2. A number of cells for I and J
M AN U
direction are defined as 100 and 150. Numbers of multi-grids and cells in the search mask are 3 and 1226, respectively. SIS lithofacies model is compared with MPS models. SIS method has been used for modeling about discrete factor with variogram modeling. When variogram modeling has been conducted for each lithofacies, the experimental variograms of all facies are fit with spherical variogram model (Table 2). Their range and nugget of each facies are
TE D
calculated by experimental variogram model (Table 2). Figs. 6 and 7 show lithofacies models from each TI. Non-reservoir facies, the Wabiskaw
EP
D shale bed, are laterally continuous and predominantly distributed on the upper part regardless of TIs (Fig. 6a–c). Reservoir facies generally is distributed on the lower part in the
AC C
western area (Fig. 7a–c). The distribution patterns show that these non-reservoir facies would act as a cap rock to the reservoir facies (Hein and Cotterll, 2006). Shaly sand facies of the Orinoco model show a parallel continuity in the eastern parts. Contrastively, shaly sand facies in the Salmon and Danube models are partially distributed. The distribution of non-reservoir facies is similar in SIS and MPS models (Fig. 6), while reservoir facies in SIS model are scattered without any continuity in the eastern part (Fig. 7d). Shale facies is dominantly distributed on the contrast sand facies is less distributed
11
ACCEPTED MANUSCRIPT in the eastern part in study area (Jo and Ha, 2013), which is well represented by the Orinoco and Salmon models rather than the Danube and SIS models. 3.3. Blind-well test for validation
RI PT
Zonal blind-well test is conducted in order to measure sensitivity of TI transitional subenvironments from sand to shale dominant settings (Fig. 2). In the transitional zone (12 and 13 sections, township 76 – range 7, Fig. 2), estuary sand dominant changes marine shale
SC
dominant from west to east laterally. In this test, control group is located in the center of
M AN U
subenvironments (Fig. 2). These zones are less shown of lateral lithofacies change (Fig. 2). The test is repeated ten times for each zone (Table 3). Similarities are calculated between simulated lithofacies and hard data blinded. If lithofacies corresponds with hard data concealed, similarity is 1 else 0. Average and variance of similarity are calculated in Table 3. For the Orinoco model, averages of blind-well test for transitional subenvironment suddenly
TE D
decreased down to 32% in Table 3 whereas averages of similarity for the center of subenvironments remain over 50% (Table 3). Blind-well test for transitional zone, the Salmon model shows the highest similarity average (Table 3), meaning that the Salmon
EP
model is the most approximate to realize distribution of sand facies in the Wabiskaw Member.
AC C
The MPS models for environmentally transitional zone can be optimized and improved by modeling in separated regions, reflecting subenvironments, with suitable TIs to each region. To compare lithofacies models from MPS and SIS, six-blind-well tests are conducted. To compare lithofacies models from MPS and SIS Table 4 shows that the MPS method is more accurate in generating lithofacies model than SIS method. Also, distribution of lithofacies model by SIS has a limitation of facies continuity (Figs. 6 and 7d). From these reasons, proper MPS modeling can reduce uncertainty of lithofacies model, however for the area with insufficient data, SIS method is still more useful than MPS to simulate lithofacies model 12
ACCEPTED MANUSCRIPT (Table 5). 4. NPV volume generation In-situ recovery such as SAGD and cyclic steam stimulation produces half of the Canadian
RI PT
bitumen production (CAPP, 2015). The SAGD was commercialized in the 1990s and has been widely used in Athabasca oilsands region where study area is located. NPV properties of the Wabiskaw Member are estimated. To evaluate the SAGD potential in study area, the
SC
Salmon lithofacies are calculated by reservoir properties such as porosity, permeability and
4.1. Reservoir property modeling
M AN U
saturation.
Rock cores of 13 boreholes and 201 well-logs are used to build porosity volume. Well-log porosity is calculated using neutron-density porosity equation with shale volume. The
TE D
calculated porosity is used to verify effectiveness to apply to lithofacies model, therefore well-log porosity is calibrated with core porosity. The vertical variations of calculated porosity from well-log are similar to measured core porosity. Spatial distribution realization
EP
of porosity is simulated using SGS method. Fig. 8a shows three dimensional distributions of porosity of the Salmon MPS model. The porosity more than 30 % in the Salmon model is
AC C
dominantly developed in the western part (Fig. 8a). These aspects are likely to distribution of reservoir facies (Fig. 7b).
Relationship of porosity and horizontal permeability for each reservoir facies is the normal regression. Permeability value is calculated from porosity using experiential regression function. Spatial distribution of permeability corresponded to porosity and permeability is zero in the region where non-reservoir facies exist in the cells. In the Salmon, permeable reservoirs (over than 1000 md) are mainly distributed in the western part (Fig. 8b). 13
ACCEPTED MANUSCRIPT In this study, water saturation is calculated from well-logs by Archie’s equation (Archie, 1942) with Pickett-plot (Pickett, 1966), supposing most of reservoir have good porosity and are relatively free from shale effect. Spatial distribution of water saturation is calculated using
RI PT
the Kriging method with porosity volume as a secondary value. Oil saturation volume can be generated from water saturation (Fig. 8c). The oil saturated area (more than 30% in pore) in the Salmon model is mostly accumulated in western part of the study area (Fig. 8c).
SC
4.2. NPV volume generation
M AN U
Generally, economy of reservoir has been evaluated with production data, it is therefore difficult to find economic sweetspot in the reservoir of the frontier or undeveloped lease. SPIDER by Shin (2008) was designed as equation of reservoir parameters; vertical thickness (H), vertical permeability (Kv), oil saturation (So), porosity (ø), SORE.L (economical limit of steam and oil ratio), and crude bitumen price (Poil). This equation is modified and applied to
(Shin, 2008).
H K S φ SOR E.L P ) × ( v )0.8 × ( o )3 × ( )1.5 × ( ) × ( oil ) 20 1 0.5 0.3 4 20
AC C
EP
SPIDER = (
TE D
build 3D economic volume and to evaluate economic of the undeveloped Wabiskaw Member
The vertical thickness (H) is 1 m on basis of vertical spacing of grid. The value of each reservoir parameter is derived from the static models. According to Meyer and Karuse (2006), the ratio between vertical and horizontal permeability is about 0.72 in the subtidal to intertidal sandbar deposits in outer estuary environment, which is used to calculate vertical permeability. The SORE.L is fixed parameter as 4 and the crude bitumen price (Poil) ranges from US$ 20 to 40.
14
ACCEPTED MANUSCRIPT The SPIDER and economic property (NPV) are also high correlation (Shin, 2008). Original SPIDER was developed to SAGD project scale with 100 × 900 m in dimension which is rescaled to static model dimension, 5 × 5 m.
NPV($) =
RI PT
SPIDER = 0.26 NPV($) + 2.2 SPIDER − 2.27 5 × 5 × Million$ × 0.26 100 × 900
SC
Hv K S φ SOR E.L P ) × ( v )0.8 × ( o )3 × ( ) × ( ) × ( oil ) − 2.27 1 1 0.5 0.3 4 20 NPV ($) = 20 × × M$ 0.26 3600
M AN U
(
The reservoir parameter values and economic parameters in each cell are used to simulate the Salmon NPV volumes according to bitumen price changing US$ 20 to 40. All cells, having NPV over than US$ 10,000 in cell, are considered as promising spots in this study.
TE D
Good horizontal continuity and concentrated appearance are shown on the Fig. 8d. The developable subjects are mainly distributed in western bottom part in the study area (Fig. 8d), if crude bitumen price is US$ 40. The ranges of porosity and permeability in promising area
EP
are shown to commonly known value in porous and permeable reservoir (Table 6). Also, oil
AC C
saturation in valuable area is over than 27% in grid (Table 6). Process through the NPV volume to evaluate economy for oilsands reservoir saves time and reduces cost, because this process only uses a few reservoir property to calculate NPV without reservoir simulation. Because of above reason, it can be applied to undeveloped or absent dynamic simulation data in oilsands lease. The volume can be utilized selection of promising area in oilsands reservoir with cut-off simply. Other factors, which affect to oil and gas production process, are not considered in this study, so NPV in grid is depending on reservoir character. 15
ACCEPTED MANUSCRIPT 5. Discussion Lithofacies modeling for complex reservoir has been challenged by using several kinds of geostatistical methods or optimizing modeling workflow (Deutsch and Journel, 1992; Yao,
RI PT
2002; Deutsch, 2006). The most-commonly-used method, SIS, has a limitation on realization of complex reservoir (Remy et al., 2009), which is improved by OBM method. The OBM used reservoir bodies with discrete shapes in 3D space and is successful to mimic complex
SC
reservoir such as fluvial channels (Deutsch and Tran, 2002; Fustic et al., 2013). But it needs to parameterize the specific geological object (Hassanpour et al., 2013) and dense well data
M AN U
may cause long simulation run time or artifacts (Pyrcz and Deutsch, 2014). On the other hand, texture-based method developed by Guardiano and Srivastava (1993) overcomes limitation of SIS and OBM, and then Strebelle (2000) suggested the procedure of obtaining probability from TI using search tree. The MPS simulation introduced geological information to static
TE D
model by training image (Strebelle, 2002), however some restrictions related to training image have newly arisen: selection of training image, proper size of training image and zonation about transitional environment for training image in vertical and horizontal
EP
directions. Verification for training image plays an important role to build accurate static model and further process, which can be commonly done by history matching (Hoffman et al.,
AC C
2006; Hoffman and Caers, 2007; Gardet et al., 2014). Because the studied reservoir is a still frontier where a production has not been started, extensive blind well test is applied to verify MPS model instead of history matching. The zonal blind-well tests highlight importance of proper zonation in MPS modeling when hard data (well data) are not enough to cover entire area. The similarity of MPS models decreases abruptly from 61% to 37.8% when sections 12 and 13 in the township 076- range 07 (T76-R07) are hidden, whereas doesn’t sections 2 and 11 in T76-R7 (57.9%), 10 and 11 in
16
ACCEPTED MANUSCRIPT T76-R6 (57.1%), 14 and 23 in T76-R7 (59.9%), and 9 and 10 in T76-R6 (57.8%) (averages of MPS models in Table 3). Sections 12 and 13 in T76-R7 are located in facies transitional zone due to change of depositional environments. In case that well data couldn’t cover facies
RI PT
transitional zone, it is better to regionalize this zone separately into sand-dominant and shaledominant parts. From iterative six-blind-well tests, Salmon MPS model shows the highest similarity (61.6%) and precision (standard deviation, 0.06) among MPS and SIS models,
SC
implying differences in geological information have effect on static modeling, although the tests range from 55.9% to 61.6% in similarity. The results of MPS modeling have almost
M AN U
same similarity about 60% to 61% but standard deviations show wider ranges, indicating Salmon MPS model will have precise prediction rather than others. Based on extensive blind well tests, a proper zonation (regionalization) is more sensitive to acquire reliable static models than a selecting proper training image in studied reservoir.
TE D
Additionally, NPV volume is generated from relationship between reservoir properties and SAGD production (Shin, 2008). This method simply shows 3D distribution of expected profits from the cell by SAGD production. However, the NPV volume doesn’t provide
EP
optimized well trajectory and ultimate recovery of SAGD project but confirm economical area and horizon of frontier reservoir. In studied area, sections 2, 3, 10, 11, 14, 15, 22, and 23
AC C
in T76-R7 are promising for SAGD project, but detailed dynamic simulation is still necessary. Vertical connectivity of bitumen reservoir is the most delicate factor for successful SAGD production, i.e. inclined heterolithic stratification and mud baffles (Gates and Larter, 2014; Shinn et al., 2014), which is not concerned in NPV volume. Moreover, commercial total thickness of bitumen sandstone should be more than 12 to 15 m (Shin, 2008), which also cannot be evaluated by the volume. Although dynamic simulation for promising area should be accomplished for exact evaluation and production design, NPV volume for frontier
17
ACCEPTED MANUSCRIPT reservoir can decrease simulation area and time enormously. 6. Conclusion The Wabiskaw D beds in this study area are simulated to lithofacies models by using MPS
RI PT
and SIS. Modeling results are shown as the considerably similar distribution aspect of lithofacies, however, horizontal continuity and detail distribution of lithofacies have difference among models. From two types blind-well tests, the Salmon lithofacies model is
SC
well represented of depositional environment of the Wabiskaw Member. However SIS model
M AN U
is more reliable than MPS models in the environment transitional zone in the case without hard data.
The static models of reservoir parameters are based on the Salmon lithofacies model. These parameters are calculated from log data and measured core data. In this study, the
TE D
SPIDER equation applies to the Wabiskaw D beds and to find economic area where located in western bottom part. Ranges of reservoir parameters in the cell over than US$ 10,000 are that porosity, horizontal permeability, and oil saturation are 29% to 43%, 2.26 to 15.42 Darcy,
AC C
EP
and 27% to 94%.
18
ACCEPTED MANUSCRIPT
Acknowledgement This work was supported by the Energy Efficiency and Resources Program of the Korea of
Energy
Technology
Evaluation
and
Planning
(KETEP,
Grant
No.
RI PT
Institute
20132510100060) and the Basic Research Program of Korea Institute of Geoscience and Mineral Resources (KIGAM, GP2015-034), funded by the Ministry of Science, ICT and
SC
Future Planning of Korea. Special thanks to Schlumberger Corporation for supporting the petrel software. Dr. Shaji E. and anonymous reviewers are warmly thanked for their helpful
AC C
EP
TE D
M AN U
review of the manuscript.
19
ACCEPTED MANUSCRIPT
Reference Alberta Energy and Utilities Board (AEUB), 2003. Athabasca Wabiskaw-McMurray Regional Geological Study. Alberta Energy and Utilities Board Report, 22-28 pp.
RI PT
Archie, G.E., 1942. The electrical resistivity log as an aid in determining some reservoir characteristics. Transactions of the AIME 146(01), 54-62.
SC
Barson, D., Bachu, S., Esslinger, P., 2001. Flow systems in the Mannville Group in the east-central Athabasca area and implications for steam-assisted gravity drainage (SAGD) operations for in situ bitumen production. Bulletin of Canadian Petroleum Geology 49(3), 376-392.
M AN U
Benyon, C., Leier, A., Leckie, D.A., Webb, A., Hubbard, S.M., Gehrels, G., 2014. Provenance of the Cretaceous Athabasca Oil Sands, Canada: Implications for continental-scale sediment transport. Journal of Sedimentary Research 84(2), 136-143. Caers, J., Zhang, T., 2004. Multiple-point geostatistics: a quantitative vehicle for integrating geology analogs into multiple reservoir models. AAPG Memoir 80, 383394.
TE D
Canadian Association of Petroleum Producers (CAPP), 2015. Crude oil forecase, markets, and transportation. Canadian Association of Petroleum Producers, Canada’s Oil and Natural Gas Producers, 2015-0007, 6-8 pp. Carrigy, M.A., 1967. Some sedimentary features of the Athabasca oil sands. Sedimentary Geology 1, 327-352.
EP
Dalrymple, R.W., Makino, Y., Zailtlin, B.A., 1991. Temporal and spatial patterns of rhythmite deposition on mud flats in the macrotidal Cobequid Bay-Salmon River estuary. Bay of Fundy, Canada.
AC C
Dalrymple, R.W., Choi, K., 2007, Morphologic and facies trends through the fluvial– marine transition in tide-dominated depositional systems: A schematic framework for environmental and sequence-stratigraphic interpretation. Earth-Science Reviews 81(3), 135-174. de Carbalho, P.R.M., da Coast, J.F.L., Rasera, L.G., Varella, L.E.S., 2016. Geostatistical facies simulation with geometric patterns of a petroleum reservoir. Stochastic Environmental Research and Risk Assessment, 1-88. Deutsch, C.V., Journel, A.G., 1992. Geostatistical software library and user’s guide. New York, 119, 147 pp. Deutsch, C.V., Tran, T.T., 2002. FLUVSIM: a program for object-based stochastic modeling of fluvial depositional systems. Computers and Geosciences 28(4), 525-535. Deutsch, C.V., 2006. A sequential indicator simulation program for categorical variables with point and block data: Block SIS. Computers and Geoscience 32(10), 1669-1681. 20
ACCEPTED MANUSCRIPT Dubois, M.K., Bohling, G.C., Chakrabarti, S., 2007. Comparison of four approaches to a rock facies classification problem. Computers and Geosciences 33(5), 599-617. Flach, P.D., 1984. Oil Sands Geology-Athabasca Deposit North. Bulletin 46, Alberta Research Counsil, Edmonton, 5.
RI PT
Fustic, M., Thurston, D., Al-Dliwe, A., Leckie, D.A., Cadiou, D., 2013. Reservoir modeling by constraining stochastic simulation to deterministically interpreted threedimensional geobodies: Case study from Lower Cretaceous McMurray Formation, Long Lake steam-assisted gravity drainage project, Northeast Alberta, Canada. AAPG Studies in Geology 64, 565-604.
SC
Gardet, C., Le Ravalec, M., Gloaguen, E., 2014. Multiscale parameterization of petrophysical properties for efficient history-matching. Mathematical Geosciences 46(3), 315-336.
M AN U
Gates, I.D., Larter, S.R., 2014. Energy efficiency and emissions intensity of SAGD. Fuel 115, 706-713. Giosan, L., Donnelly, J.P., Vespremeanu, E., Bhattacharya, J.P., Olariu, C., Buonaiuto F.S., 2005. River delta morphodynamics: examples from the danube delta. SEPM Special Publication 83, 393-411. Guardiano, F., Srivastava, M., 1993. Multivariate geostatistics: beyond bivariate moments. Geostatistics-Toria 92, 133-144.
TE D
Jo, H.R., Ha, C.G., 2013, Stratigraphic architecture of fluvial deposits of the Cretaceous McMurray Formation, Athabasca oil sands, Alberta, Canada. Geosciences Journal 17(4), 417-427.
EP
Jung, A., Aigner, T., Palermo, D., Nardon, S., Pontiggia, M., 2012, A new workflow for carbonate reservoir modeling based on MPS: shoal bodies in outcrop analogue(Trissic, SW Germany). Geological Society, London, Special Publication 370(1), 277-293.
AC C
Hashemi, S., Javaherian, A., Ataee-pour, M., Tahmasebi, P., Khoshdel, H., 2014, Channel characterization using multiple-point geostatistics, neural network, and modern analogy: A case study from a carbonate reservoir, southwest Iran. Journal of Applied Geophysics 111, 47-58. Hassanpour, M.M., Pyrcz, M.J., Deutsch, C.V., 2013, Improved geostatistical models of inclined heterolithic strata for McMurray Formation, Alberta, Canada. American Association of Petroleum Geologists Bulletin 97(7), 1209-1224. Hein, F.J., Cotterill, D.K., 2006, The Athabasca oil sands-a regional geological perspective, Fort McMurray area, Alberta, Canada. Natural Resources Research 15(2), 85-102. Hoffman, B.T., Caers, J.K., Wen, X.H., Strebelle, S., 2006. A practical data integration approach to history matching: application to a deepwater reservoir. SPE Journal 11(04), 464-479.
21
ACCEPTED MANUSCRIPT Hoffman, B.T., Caers, J., 2007, History matching by jointly perturbing local facies proportions and their spatial distribution: Application to a North Sea reservoir. Journal of Petroleum Science and Engineering 57(3), 257-272.
RI PT
Keith, D.A.W., Wightman, D.M., Pemberton, S.G., MacGillivray, J.R., Berezniuk, T., Berhane, H., 1988. Sedimentology of the McMurray Formation and Wabiskaw Member (Clearwater Formation), Lower Cretaceous, in the central region of the Athabasca oil sands area, northeastern Alberta. Sequence stratigraphy, sedimentology: surface and subsurface: Canadian Society of Petroleum Geologist, Memoir 15, 309324.
SC
Leckie, D.A., Smith, D.G., 1992, Regional setting, evolution and depositional cycles of the Western Canada Foreland Basin. Foreland Basins and Fold Belts: American Association of Petroleum Geologists, Memoir 55, 9-46.
M AN U
Lee, K., Jeong, H., Jung, S., Choe, J., 2013. Improvement of ensemble smoother with clustering covariance for channelized reservoirs. Energy Exploration and Exploitation 31 (5), 713-726. Lim, J., Jin, J., Choe, J., 2014. Features Modeling of Oil Sands Reservoirs in Metrics Space. Energy Sources, Part A: Recovery, Utilization and Environmental Effects 36(24), 2725-2735. Maharaja, A., 2008. TIGenerator: Object-based training image generator. Computers and Geosciences 34(12), 1753-1761.
TE D
Meyer, R., Krause, F.F., 2006. Permeability anisotropy and heterogeneity of a sandstone reservoir analogue: An estuarine to shoreface depositional system in the Virgelle Member, Milk River Formation, Writing-on-Stone Provincial Park, southern Alberta. Bulletin of Canadian Petroleum Geology 54(4), 301-318.
EP
Negi, J.K., Verma, C.P., Kumar, A., Prasad, U.S., and Lal, C., 2006. Predicting lithofacies using artificial neural network and log-core correlation. In: Proceedings of 6th international Conference and Exposition on Petroleum Geophysics, 809-811.
AC C
Panin, N., Jipa, D., 2002. Danube river sediment input and its interaction with the northwestern Black Sea. Estuarine, Coastal and Shelf Science 54(3), 551-562. Pickett, G.R., 1966. A review of current techniques for determination of water saturation from logs. Journal of Petroleum Technology 18(11), 1425-1433. Pyrcz, M.J., Boisvert, J.B., Deutsch, C.V., 2008. A library of training images for fluvial and deepwater reservoirs and associated code. Computers and Geosciences 34(5), 542560. Pyrcz, M.J., Deutsch, C.V., 2014. Geostatistical reservoir modeling. Oxford University Press, 249-266 pp. Qi, L., Carr, T.R., 2006. Neural network prediction of carbonate lithofacies from well logs, Big Bow and Sand Arroyo Creek fields, Southwest Kansas. Computers and Geosciences 32(7), 947-964. 22
ACCEPTED MANUSCRIPT Remy, N., Alexandre, B., Jianbing, W., 2009. Applied geostatistics with SGemS: A user’s guide. Cambridge University Press, New York. Shields, D.J., Strobl, R., 2010. The Wabiskaw D Member, Clearwater Formation: A world class Oil Sands reservoir hosted in an incised valley complex. GeoCanada 2010Working with the Earth.
RI PT
Shin, H., 2008. A Reservoir economic indicator to evaluate SAGD project. The Canadian f International Petroleum Conference/SPE Gas Technology Symposium 2008, PAPER 2008-072.
SC
Shinn, Y.J., Lee, H.S., Kwon, Y.K., Kwak, W.J., 2014. Lithofacies distribution and depositional environment in the Lower Cretaceous McMurray Formation, BlackGold Lease, northern Alberta: implications for geometry and distribution of oil sand reservoirs. Geosciences Journal 18(3), 325-337.
M AN U
Strebelle, S., 2000. Sequential simulation drawing structures from training images. Ph.D. Thesis, Department of Geological and Environmental Sciences, Stanford University. Strebelle, S., 2002. Conditional simulation of complex geological structures using multiplepoint statistics. Mathematical Geology 34(1), 1-21. Wang, G., Carr, T.R., 2012. Methodology of organic-rich shale lithofacies identification and prediction: A case study from Marcellus Shale in the Appalanchian basin. Computers and Geosciences 49, 151-163.
TE D
Warne, A.G., Meade, R.H., White, W.A., Guevara, E.H., Gibeaut, J., Smyth, R.C., Aslan, A., Termblay, T., 2002, Regional controls on geomorphology, hydrology, and ecosystem integrity in the Orinoco Delta, Venezuela. Geomorphology 44, 273-307.
EP
Wightman, D.M., Strobl, R.S., Cotterill, D.K., Berhane, H., Attalla, M.N., 1997. Stratigraphy, depositional modelling and resource characterization of the McMurray/Wabiskaw Deposit, western portion of the Athabasca Oil Sands area, northeastern Alberta. Petroleum geology of the Cretaceous Mannville Group, Western Canada, Canadian Society of Petroleum Geologists, Memoir 18, 345-374.
AC C
Wong, P.M., Jian, F.X., Taggart, I.J., 1995. A critical comparison of neural networks and discriminant analysis in lithofacies, porosity and permeability predictions. Journal of Petroleum Geology 18(2), 191-206. Wynne, D.A., Attalla, M., Berhane, H., Cotterill, D.K., Strobl, R., Whiteman, D.M., 1994. Athabasca oil sands database: McMurray/Wabiskaw Deposit. AEUB/AGS Report, 1994-14, 14-17 pp. Yao, T., 2002. Integrating seismic data for lithofacies modeling: A comparison of sequential indicator simulation algorithms. Mathematical Geology 34(4), 387-403.
23
ACCEPTED MANUSCRIPT
Table 2 Settings to simulate the lithofacies for MPS and SIS.
RI PT
Table 1 Sub-environment, porosity, permeability, and oil saturation related to facies association (FA) for the Wabiskaw Member in the Kirby North Lease (modified from Shields and Strobl, 2010).
Table 3 Average and variance for similarities of iterative of zonal blind-well tests (zonation with symbol displayed on Fig. 3).
SC
Table 4 Similarities of iterative random six-blind-wells test from MPS and SIS modeling results.
AC C
EP
TE D
M AN U
Table 5 Similarities of iterative zonal blind-well tests for SIS modeling.
24
ACCEPTED MANUSCRIPT Table 6 Ranges of porosity, horizontal permeability, and oil saturation according to change of crude bitumen price for the Salmon NPV model (economic criterion in a cell: US$ 10,000).
RI PT
Figure 1. Generalized paleogeography map during the Early Cretaceous. Latitude and longitude on the map show present coordinate. Black arrows, gray, and white area represent paleocurrent direction, high relief region, and area of deposition, respectively (modifed from Leckie and Smith, 1992; Benyon et al., 2014). Figure 2. Location of boreholes in the study area (top view) and indication of zonal blindwell test. Blue and yellow symbols are located in the estuary to shallow marine depositional environment and purple and green symbols are included in the shallow marine depositional environment. Red symbols stand for environmental transition zone.
M AN U
SC
Figure 3. (a) Satellite image of the Orinoco River Delta estuary, consisting of river dominated (upper) and tide dominated (lower) sections. (b) Location of TI in the Orinoco River Delta estuary satellite image. (c) Selected TI in the Orinoco River Delta estuary environment, and (d) generated TI with four lithofacies. Figure 4. (a) Satellite image of the Cobequid bay-Salmon River estuary. Selected TI in red box on (b) for the Cobequid bay-Salmon River esturay (Dalrymple and Choi, 2007). (c) Selected TI on red box from the (a). (d) Generated TI with four lithofacies.
TE D
Figure 5. (a) Satellite image of the Danube River delta, red box indicating the location of selected TI. (b) Image of TI of the Danube River delta, red arrows indicating the current direction. (c) Generated TI with four lithofacies.
EP
Figure 6. Non-reservoir facies distribution of MPS and SIS lithofacies models (vertical exaggeration X 20).
AC C
Figure 7. Reservoir facies distribution of MPS and SIS lithofacies models (vertical exaggeration X 20). Figure 8. Distribution of Salmon petrophysical properties and NPV (vertical exaggeration X 20).
25
ACCEPTED MANUSCRIPT Table 1. Sub-environment, porosity, permeability, and oil saturation related to facies association (FA) for the Wabiskaw Member in the Kirby North Lease (modified from Shields and Strobl, 2010). Porosity
Permeability
FA1
Tidal influenced channel fill deposits
30–33%
Around 1 Darcy
80%
FA2 FA3
High energy bay fill deposits Shoreface succession
26–29% -
