You are on page 1of 27

Analysis and modeling of

intermediate-scale reservoir
heterogeneity based on a fluvial
point-bar outcrop analog,
Williams Fork Formation,
Piceance Basin, Colorado
Matthew J. Pranter, Amanda I. Ellison, Rex D. Cole, and
Penny E. Patterson

ABSTRACT
This study presents results of outcrop characterization and modeling of lithologic heterogeneity within a well-exposed point bar of
the Williams Fork Formation in Coal Canyon, Piceance Basin, Colorado. This deposit represents an intermediate-scale depositional
element that developed from a single meandering channel within a
low net-to-gross ratio fluvial system. Williams Fork outcrops are analogs to petroleum reservoirs in the Piceance Basin and elsewhere.
Analysis and modeling of the point bar involved outcrop measurements and ground-based high-resolution light detection and ranging
data; thus, the stratigraphic frameworks accurately represent the
channel-fill architecture.
Two- and three-dimensional (2-D and 3-D) outcrop models and
streamline simulations compare scenarios based on different lithologies, shale drapes, observed grain-size trends, petrophysical properties, and modeling methods. For 2-D models, continuous and discontinuous shale drapes on lateral-accretion surfaces result in a 79%
increase and 24% decrease in breakthrough time (BTT), respectively, compared to models without shale drapes. The discontinuous
shale drapes in the 2-D and 3-D models cause a 30% and 107%
decrease, respectively, in sweep efficiency because they focus fluid
flow downward to the base of the point bar. For similar reasons, 2-D
models based on grain size exhibit 67–267% shorter BTT and 44–
57% lower sweep efficiency compared to other model scenarios.
Unlike the 2-D models, the continuous shale drapes in the 3-D

Copyright #2007. The American Association of Petroleum Geologists. All rights reserved.
Manuscript received August 31, 2006; provisional acceptance November 7, 2006; revised manuscript
received January 19, 2007; final acceptance February 1, 2007.
DOI:10.1306/02010706102

AAPG Bulletin, v. 91, no. 7 (July 2007), pp. 1025 –1051

1025

AUTHORS
Matthew J. Pranter  Department of
Geological Sciences and Energy and Minerals
Applied Research Center, University of Colorado, UCB 399, Boulder, Colorado 80309;
matthew.pranter@colorado.edu
Matt Pranter is an assistant professor at the
University of Colorado at Boulder and head of
the Reservoir Characterization and Modeling
Laboratory. He received his B.S. degrees in
geology and geological engineering from Oklahoma State University and the Colorado School
of Mines, respectively, his M.S. degree in geology from Baylor University, and his Ph.D. in
geology from the Colorado School of Mines. He
was previously with ExxonMobil Upstream
Research Company and Conoco. His research
interests are in reservoir geology and geophysics,
sedimentary geology, and reservoir modeling.
Amanda I. Ellison  Department of Geological Sciences, University of Colorado,
UCB 399, Boulder, Colorado 80309; present
address: ExxonMobil Upstream Research
Company, 3120 Buffalo Speedway, URC-N328,
Houston, Texas 77098;
amanda.ellison@exxonmobil.com
Amanda Ellison is a geologist and geologic
modeler at ExxonMobil Upstream Research
Company. She received her B.S. degree in
geology from Colorado State University and
her M.S. degree in geology from the University
of Colorado at Boulder. Her interests are in
reservoir characterization and modeling.
Rex D. Cole  Department of Physical and
Environmental Sciences, Mesa State College,
1100 North Avenue, Grand Junction, Colorado
81501; rcole@mesastate.edu
Rex Cole is a professor of geology at Mesa
State College in Grand Junction, Colorado. He
obtained his A.S. degree in geology from Mesa
Junior College, his B.S. degree in geology from
Colorado State University, and his Ph.D. in
geology from the University of Utah. Previous
employers include Unocal Corporation, MultiMineral Corporation, Bendix Field Engineering Corporation, Southern Illinois University –
Carbondale, and Asarco Corporation.

Penny E. Patterson  ExxonMobil Exploration Company, 233 Benmar, Houston, Texas
77060; penny.e.patterson@exxonmobil.com
Penny Patterson is a geological associate at
ExxonMobil Exploration Company. She received
her B.A. and M.S. degrees in geology at the
University of Colorado. Penny then worked for
the Research Planning Institute, specializing
in nearshore marine, fluvial, and eolian strata.
She earned her Ph.D. in fluvial sedimentology
and stratigraphy and sandstone diagenesis at
the University of Colorado. She then joined
Exxon Production Research Company, where her
research focused on nonmarine and shallowmarine stratal architecture and sequencestratigraphic concepts. Currently, she is involved
in international exploration opportunities at
ExxonMobil Exploration Company.

ACKNOWLEDGEMENTS
Financial support of this research was provided
through the Reservoir Characterization and
Modeling Laboratory at the University of Colorado at Boulder, the AAPG Foundation (AAPG
Grants-In-Aid Program), Rocky Mountain Association of Geologists, and the International
Association of Mathematical Geologists. We
thank ExxonMobil Upstream Research Company for funding the acquisition of the groundbased light detection and ranging data for this
study. Dave Jennette and Florence Bonnaffe
of the Texas Bureau of Economic Geology provided assistance with LIDAR processing and
interpretation. Neil Hurley of the Colorado
School of Mines provided a gamma-ray scintillometer for outcrop gamma-ray measurements.
We thank the U.S. Bureau of Land Management
for providing access to Coal Canyon. Quentin
German and Tim Smith provided assistance in
the field. We thank Andrew Miall, Richard
Moiola, Brian Horn, Colin North, Jim Rogers, and
Art Trevena for their reviews of this article.

1026

models cause the fluid front to spread out and contact more of the
reservoir, resulting in 42–53% longer BTT and 41–52% higher sweep
efficiency compared to the other models. These results provide additional insight into the significance of intermediate-scale heterogeneity of fluvial reservoirs.

INTRODUCTION
Characterization and modeling of fluvial reservoirs are challenging
because of the various scales of heterogeneity that exist between
and within fluvial deposits (Jackson, 1977; Miall, 1988; Willis, 1989;
Sharp et al., 2003). Field-scale heterogeneity (e.g., heterogeneity
levels 1 and 2 of Jordan and Pryor, 1992) of a fluvial reservoir is dependent on the hierarchy of architectural elements, sand body stacking patterns, and associated connectivity between reservoir sand bodies
(Robinson and McCabe, 1997; Anderson, 2005). Intermediate-scale
heterogeneity within individual fluvial sand bodies (e.g., heterogeneity levels 3 and 4 of Jordan and Pryor, 1992) also affects reservoir
performance and is generally associated with facies distribution, grain
size and sorting, and lithology variation (Richardson et al., 1978;
Lasseter et al., 1986; Thomas et al., 1987; Tyler and Finley, 1991;
Hartkamp-Bakker and Donselaar, 1993; Willis and White, 2000).
The large-scale stratigraphic architecture and intermediate-scale internal heterolithic stratification of fluvial reservoirs are commonly
difficult to characterize with subsurface data because of the relatively limited lateral extent of fluvial deposits, which are commonly
less than the typical well spacing in developed fields (e.g., 10–40-ac
[201–402-m; 660–1320-ft] spacing). In the case of isolated pointbar reservoirs of low net-to-gross (low sandstone-to-shale) ratio
systems, common internal depositional features are lateral-accretion
units and associated bedding that create intermediate-scale variability. These depositional units and associated shale drapes can influence reservoir behavior (Swanson, 1993) because they are potential
baffles and barriers to fluid flow (Richardson et al., 1978; HartkampBakker and Donselaar, 1993). In addition, internal facies variations,
such as the vertical change from trough cross-bedded to ripplelaminated sandstone, produce a fining-upward trend in grain size
and a corresponding decrease in porosity and permeability.
To address these issues, outcrop analogs of well-exposed fluvial
deposits of the Williams Fork Formation were evaluated and modeled. The outcrops are located in Coal Canyon near Palisade, Colorado, and are approximately 48 km (30 mi) southwest of natural-gas
fields in the southern Piceance Basin that produce from the Williams
Fork Formation (Figures 1, 2). In addition to its display of largescale sand body heterogeneity (i.e., connectivity between sand bodies at a field scale), Coal Canyon is an excellent location to study
intermediate-scale sand body heterogeneity (i.e., within a fluvial sand
body) because vegetation is minimal, rocks are well exposed, structural complications are minimal (structural dips are generally < 7j),
and access is good.

Outcrop Modeling of Point-Bar Heterogeneity, Williams Fork Formation

which extended from the Gulf of Mexico to the Pranter et al. Detailed two. GEOLOGIC AND STRATIGRAPHIC SETTING The Piceance Basin is an elongate northwest-southeast– trending downwarp created during the Laramide orogeny (Late Cretaceous–early Paleogene). the Piceance Basin area was situated on the western foreland margin of the Western Interior seaway. Therefore. photomosaics. there are common characteristics that link them. Several depositional scenarios and modeling methods are used to model lithology and petrophysical properties and evaluate the relative effects of internal heterogeneities on fluid-flow behavior. The point bar of this study is exposed through three primary outcrops that provide cross sectional views of the deposit in both depositional-dip and depositional-strike orientations. Location map of the Piceance Basin and study area. The study area includes outcrops of pointbar deposits at a locality within Coal Canyon located on the western basin margin north of Palisade. Outcrops of the Mesaverde Group occur on the margins of the basin (dark-gray shading). 1027 . Although fluvial reservoirs are diverse and unique.and three-dimensional (2-D and 3-D) outcrop models of a well-exposed point-bar deposit (single reservoir element or sand body) in the lower Williams Fork Formation (Campanian) of the Mesaverde Group of western Colorado were constructed using highresolution ground-based light detection and ranging (LIDAR) data (digital laser images). we believe that the results of this study are applicable to a variety of point-bar reservoirs. Miall. 1996). 1963. such as the origin of the depositional system and the general geometry of the reservoir bodies (Schumm. and field measurements.Figure 1. During the Late Cretaceous. Modified from Hoak and Klawitter (1997).

The Williams Fork Formation thins from approximately 1525 m (5003 ft) on the eastern side of the Piceance Basin to about 370 m (1213 ft) at the Colorado – Utah border (Hettinger and Kirschbaum. 2003). 2003. The middle to upper Campanian Williams Fork Formation of the Mesaverde Group conformably overlies the Iles Formation and is overlain disconformably by the Paleocene Wasatch Formation. Arctic. Geological Survey. The latitude and longitude of each outcrop are northwest outcrop (41j55005. Colorado. transported eastward by fluvial depositional systems. we follow the stratigraphic terminology of Hettinger and Kirschbaum (2002. Johnson and Flores. 104j48048. and Wyoming. 2005). Colorado. 7. and the surrounding area. (A) Topographic map of Coal Canyon. Tyler et al. Utah. Williams Fork Formation .1300W). 2003) for 1028 the Mesaverde Group in the southwestern Piceance Basin (Figure 3). southwest outcrop (41j5507.6000W).S.4000N. and east outcrop (41j5501. In the southwestern Piceance Basin. sediment that formed the Williams Fork Formation was shed from the Sevier orogenic belt in Arizona. 104j48045. The contour interval is 12 m (40 ft). 104j48040.Figure 2. the Williams Fork is subdivided Outcrop Modeling of Point-Bar Heterogeneity.. (B) Locations of the three main outcrops that were evaluated and measuredsection locations (MS 1 to MS 13). The specific study area (small dashed square) is shown in greater detail in (B). 2002.0800W). 1989. and ultimately deposited in shoreline environments that bounded the epeiric seaway (see summaries in Johnson.5-min Quadrangle (1955). 1996.2800N. The map is modified from the U. Cameo. Cole and Cumella.5800N. In this article. During this time.

to Pranter et al. currentrippled sandstone. and Upper (Cole and Cumella. and mud-chip conglomerate (rare). with subordinate lenticular channel-form sandstone and coal. Fisher et al. conglomeratic mud-chip sandstone. 5). The sand-poor deposits include relatively isolated but internally heterolithic sandstones (e. 2005). carbonaceous to very carbonaceous mudrock. and Johnson and Roberts (2003). Connectivity between the point-bar sand bodies is generally low. Sources of data include Young (1955. Six main lithofacies based on thirteen stratigraphic sections include trough cross-bedded sandstone. 2005). Highwall..g. (1996). whereas the upper 250 – 300 m (820 – 984 ft) is mostly sandstone (50–80%). (1960).Figure 3. nodular siltstone. The point-bar deposit exhibits a fining-upward succession of facies from medium. The focus of this study is on the intermediate-scale internal lithologic heterogeneity within this single fluvial point-bar sand body (analogous to a single isolated reservoir element). The Cameo-Wheeler coal zone (Figures 3. POINT-BAR SEDIMENTOLOGY Detailed sedimentologic observations and measurements were collected from three outcrops of a single point-bar and associated deposits within Coal Canyon (Figures 2. Stratigraphic nomenclature used in the Piceance Basin. swamps. ironstone (siderite) concretions. 1029 . Donnell (1961). The lower 150 – 200 m (492 – 656 ft) of the Williams Fork Formation consists mostly of mudrock (approximately 40 – 70%). The upper sand-rich interval was deposited in an alluvial-plain setting. 2005) of the lower 600 ft (182 m) of the Williams Fork Formation in Coal Canyon is illustrated in Figure 4. isolated point-bar sand bodies) encased in flood-plain mudstones and coals. Johnson and May (1980). 1966). Table 1). The outcrops described in this study are a part of the lower Williams Fork Formation in the CameoWheeler coal zone. connectivity is greater in the sand-rich intervals (high net-to-gross ratio). where amalgamated sand bodies are more common. laminated siltstone. Collins (1976). and flood plains (Cole and Cumella. 4) at the base of the Williams Fork Formation is about 73 m (240 ft) thick in Coal Canyon and consists of interstratified coal. and coal and bentonite beds (Figure 6. Ash. Four locally mappable coal seams occur in this zone: Cameo. The point-bar sand body of this investigation is located near the middle of the Cameo-Wheeler interval between the Ash and Upper coal seams (Figure 4). however. A composite section (Cole and Cumella. sandstone. and the lower sand-poor interval of the Williams Fork Formation was deposited in a coastal-plain setting with meandering streams. with subordinate mudrock and almost no coal. Tyler et al. 2003). Hettinger and Kirschbaum (2002. into two intervals based on lithofacies.

see Figure 1 for the well location). Data for the composite section are based on measured sections of Cole and Cumella (2003). Comparison of a composite section in Coal Canyon with the Dyco Petroleum 1 Sommerville well (data are modified from Hettinger and Kirschbaum. 2003. Williams Fork Formation . 1030 Outcrop Modeling of Point-Bar Heterogeneity. 2002. The study interval for the point-bar sand body of this investigation is shown on the composite log.Figure 4. The correlation line at the top of the composite section is based on thickness equivalence.

it is nearly all ripple stratified or mainly massive. Well-exposed lateral-accretion surfaces are visible on the northwest outcrop (B). Ellison. ripple-laminated sandstone (thin levee or crevasse-splay sand body) (Cole et al. The main pointbar deposit is a lens-shaped body with a sharp base and has an average thickness of 8.5 to 3 m (5 to 10 ft). 2004) (Figure 7). and lacks large-scale internal bedding.. Ellison. upper fine-grained trough cross-beds to fine-grained ripple-laminated sandstone. Pranter et al. lateral-accretion surfaces and lateral-accretion sandstone beds within the point-bar deposits have average dips between 10 and 12j. Outcrops of the lower Williams Fork Formation fluvial deposits in Coal Canyon. 2002. whereas in other sections. Cole and Cumella. 2004).0 m (26 ft). The sandstone beds are separated by lateral-accretion surfaces and mudrock interbeds and are typically cross-stratified. The crevassesplay deposit has a sheetlike geometry with a sharp planar base. The two amalgamated point-bar sand bodies are separated by an erosional surface. In some sections. Distinct high-angle. 2002. 2003. the long dimension and paleoflow direction of the main point-bar sand body is oriented to the southeast (azimuth of 94 – 102j) (Cole and Cumella. in previous work (Cole et al. 1031 . Based on lateral-accretion bedding. the main sand body is dominated by channel-form sets of trough cross-stratification. indicating that they accreted in opposite directions. The accretion surfaces dip to the south in the younger (main) point bar and to the north in the older point bar.Figure 5. (B) northwest. and (C) east outcrops. The main point-bar sand body (Figures 5. The succession is attributed to a gradual decrease in flow velocity during deposition. Cole and Cumella. this sand body was informally named the ‘‘6-50’’ sand body. 2003). 7) partially overlies an older point-bar deposit and is overlain by a thin mudstone interval (abandonment shale) and a fine-grained. 2003. For reference. Scour surfaces with mud-chip lags are locally common.. ranges in thickness from 1. The locations of 13 measured sections are shown as white vertical lines on the (A) southwest.

(A) Trough cross-bedded sandstone.Figure 6. (E) conglomeratic mud-chip sandstone. (D) laminated siltstone. Major fluvial facies of the Williams Fork Formation. 1032 Outcrop Modeling of Point-Bar Heterogeneity. (C) nodular siltstone. See Table 1 for more detailed facies descriptions. (B) current-rippled sandstone. Williams Fork Formation . and (F) coal (c) and bentonite (b) beds.

also planar cross-bedding.9 and 10 ft) Tractive flow deposit. scoured surfaces Current ripple laminations. mud drapes indicate periods of little to no ripple migration Vertical accretion. moderately sorted Nodular sandstone Mudstone Silt dominated NA* Laminated siltstone Mudstone Silt rich NA Conglomeratic mud-chip sandstone Coal and bentonite Conglomeratic lithic sandstone Coal and bentonite Upper to lower medium NA Pranter et al. moderately sorted Current rippled sandstone Lithic sandstone Upper to lower fine Subrounded. Fluvial Facies Recognized in the Coal Canyon Outcrops and Their Characteristics Facies Lithology Modal Grain Size Grain Shape and Sorting Trough cross-bedded sandstone Lithic sandstone Lower medium to upper fine Subangular to subrounded. siderite concretions. woody plant debris. suspension fallout and pedogenesis Scour and fill generated by flow separation indicates disequilibrium in flow conditions Vertical accretion processes such as suspension fallout.) Depositional Process Unidirectional tractive flow deposit. bioturbation. poorly sorted NA Dominant Features Trough cross-bedding.75 in. suspension fallout and pedogenesis Vertical accretion. continuity of coals indicates low channel activity 1033 . 0. dune height indicates water depths between 1.2 and 3 m (3. rooting Mud-chip lag. Subangular. soft sediment defined. climbing ripples indicate fluctuations in aggradation rate.Table 1. rooting Millimeter-scale laminations. *NA = not applicable.8 –2. scoured surfaces Regularly spaced cleats in coal (2 – 7 cm. climbing ripples Nodular texture.

and outcrop data of the fluvial deposits is shown in the lower panel. A cross section based on the measured sections. photomosaic. The upper bounding surface of the point-bar deposit is sharp and planar. with little to no relief. thin mudstones are common along the accretion surfaces and change downdip to silty sand1034 stones with siderite concretions. The lithologies present along the accretion surfaces vary laterally. Both point bars have well-defined lateral-accretion surfaces. The main point bar is a lens-shaped sand body defined by sharp basal-scour and top-truncated surfaces. The main point bar is separated from an older point bar by an erosional surface. The overlying levee-splay has a sheetlike geometry with a sharp base. Williams Fork Formation . Updip. Outcrop Modeling of Point-Bar Heterogeneity. Key fluvial architectural elements exposed in the northwest outcrop. The depth scales on the measured sections are in feet.Figure 7.

Because bankfull channel depth in straight reaches of a channel is less than in the meander bends. Ethridge and Schumm. Lorenz et al.1 m. 1985). both of which can be estimated from channel size (Collinson. Leopold and Wolman. outcrops provide 2-D data (e.4 m (146 ft). Therefore. b).6 m (97 ft). 1965. 1973. Meander Amplitude and Wavelength: Reservoir Lateral Dimensions The lateral dimension of the point-bar reservoir perpendicular to the paleoflow direction (reservoir width) approximately corresponds to the meander-belt width (meander amplitude) of the channel that deposited the main point bar (Collinson. channel depth. which yields a bankfull channel depth of 5. 1993a.585 for the ratio between bankfull channel depth in a straight reach versus a meander bend of a channel is used. Schumm. Therefore. For comparison. As an approximation.0 m (26 ft). potential sources of error associated with amalgamation and truncation are limited. In this situation. Pranter et al. Bridge and Mackey. The meander-belt width is estimated to be 300 m (980 ft). Based on experimental studies (Ethridge and Schumm.. Leeder. a correction factor is applied to values of point-bar thickness from outcrop or core. 1977). 1972. 1973). 17 ft).9 to convert sand body thickness measurements to original bankfull channel depth: D ¼ D  0:585=0:9 The average thickness (D*) of the main point bar studied herein (Figure 5) is approximately 8.. Ethridge and Schumm (1977) suggest that a 10% reduction in thickness for the conversion of sand to sandstone as a result of compac- tion is conservative. Allen (1965) showed that point bars commonly extend two-thirds of the distance across a channel (or two-thirds of the bankfull channel width). the horizontal widths of lateral accretion surfaces can be measured. 1978. The average horizontal width of an accretion surface (W*) in the main point-bar sand body is 29. Because the subject point bar represents a relatively isolated and complete deposit.. These empirical relationships and measurements of the main point-bar deposit in outcrop were used to estimate the paleochannel size (bankfull channel depth and width). Bankfull Channel Depth and Width The maximum lateral extent of a point-bar sand body or reservoir element relates to the meander-belt width (meander amplitude) and meander wavelength. 146 – 283 ft) is not uncommon and is expected given the different sources of data for the empirical relationships and the resulting estimates.g.e. Leeder (1973) presents an empirical relationship between bankfull channel depth and width that can be used in situations where the outcrop exposure is not adequate to measure lateral-accretion surfaces. For 3-D modeling purposes (to be discussed).4 – 86. In most cases. and meander wavelength (e.2 m (17 ft). which yields a bankfull channel width of 44.1 m (283 ft). 1978. 1977. Previously developed empirical relationships based on modern streams relate channel width. the following empirical equation is used..2 m. A range in the estimated bankfull channel width estimates (44.7). thickness and/or widths) for sandstone and shale bodies..g. or the dimensions and morphology of the paleochannel can be reconstructed or estimated from empirical data and relationships that are commonly based on modern systems (Ethridge and Schumm. Based on this. in high-sinuosity rivers (i. the plan-view geometry of the body must be inferred from modern analogs. Collinson. meander-belt width. Bankfull channel depth (D) is estimated from the average thickness of the sand body (D*) and corrected for the curvature of the meander bend and compaction (Figure 8).POINT-BAR (RESERVOIR) DIMENSIONS Outcrop analogs are commonly used to obtain dimensional statistics of fluvial sandstone or shale bodies for 3-D modeling of similar petroleum reservoirs. with an estimate of bankfull channel depth as W ¼ 6:8  D1:54 Based on this relationship and using the corrected bankfull channel depth from outcrop (5. 1978). sinuosity > 1. Lorenz et al. the estimated bankfull channel width is 86. Bankfull channel depth and width are defined as the maximum depth and width that a stream channel can attain before the discharge of floodwater into areas outside of the channel (Leeder. bankfull channel width (W ) is estimated by measuring the average horizontal width of the lateral-accretion surfaces as exposed in outcrop (W*) (Figure 8) such that W ¼ W   1:5 For the well-exposed outcrops of this study. 1035 . an average value of 0. 1977. Carlston. the point-bar thickness is also divided by 0. the plan-view dimensions and morphology of the point-bar sand body were estimated using empirical data and spatial relationships. 1985). 1960.

Given the relatively narrow range of the lowest three meander wavelength estimates. 2004). Bankfull channel depth and width are estimated from point-bar deposits in outcrop using the point-bar thickness (D*) and horizontal length of lateral-accretion surfaces (W*). and point-bar sand body dimensions parallel to the paleoflow direction exhibit a range of values relative to the meander wavelength (e.7–1. the average of meander wavelength estimates was used herein. Fluvial systems are complex.2 m [17 ft]) and bankfull channel width (W = 86. Schematic illustration of a meander bend in (A) plan view and (B) cross sectional view. 1978) (Figure 8) of a channel. based on the lateral extent of the point-bar sandstone in outcrop (i. which relate meander wavelength to bankfull channel width (W ) and bankfull channel width-to-depth ratio (F = W/D). For simplicity and given the relatively high sinuosity of the channels of the lower Williams Fork Formation (sinuosity = 1. Based on Lorenz et al. The meander wavelength cannot be directly measured from outcrop as with the meander-belt width.g. The lateral dimension of the point-bar reservoir parallel to the paleoflow direction (reservoir length) relates to the meander wavelength (Collinson. Williams Fork Formation .4 m [146 ft]) yield wavelengths of 503 and 768 m (1650 and 2520 ft). The actual point-bar lateral dimension is not exactly 750 m (2460 ft) (this is an estimate).. however. 1972) 1036 Using the estimate of bankfull channel depth (D = 5. based on Outcrop Modeling of Point-Bar Heterogeneity. 1978).. based on the observed sandstone body pinchout). the above equations yield meander wavelengths for the main point-bar deposit of 982 and 1725 m (3222 and 5660 ft). Ellison. The following empirical relationships were used to estimate meander wavelength: lm ¼ 10:9 W 1:01 ðunits ¼ mÞ (Leopold and Wolman. Collinson.1 m [283 ft]) based on the relationships of Leeder (1973).9. The meander wavelength (lm) was thus calculated using empirical relationships to estimate the lateral dimension of the point-bar reservoir parallel to the paleoflow direction. units ¼ mÞ (Schumm. Meander wavelength (Figure 8) is estimated using the methods of Leopold and Wolman (1960) and Schumm (1972).Figure 8. Geomorphic parameters include bankfull channel width (W) and meander wavelength (lm). an average value of 750 m (2460 ft) for the meander wavelength was subsequently used as the lateral dimension of the point-bar reservoir (reservoir length parallel to the paleoflow direction) for 3-D outcrop modeling. 1960) or lm ¼ 18ðF 0:53 W 0:69 Þ ðF ¼ W =D. Meander wavelength estimates based on the lower value of bankfull channel width (W = 44.e. (1985).

There have been numerous studies and methods that explore geologic modeling of various scales of heterogeneities within fluvial reservoirs. Eschard et al.. Jones et al. 1037 . Each zone is further subdivided using finescale layering to represent the stratal patterns of the accretionary bedding observed in outcrop. Cell layering was truncated at the top of each architectural element (i.. Nanson. 2000.. As a result.13  1. grain size. deterministic (Stephen and Dalrymple. Weber... The model is divided into 13 distinct zones to reflect the stratigraphic architecture of the point-bar deposit (Figure 10). MacDonald and Halland. The point bar was modeled assuming the entire point-bar sand body was preserved (Figure 11) Pranter et al.g. and sweep efficiency. Dalrymple. 1995). 1993a. Others address the potential effect of internal shale or sand body heterogeneity (lithological or petrophysical) on fluid flow (e.5 m (3.. the 2-D outcrop model framework consists of 75.. Two. Richardson et al. LIDAR data were used to interpret key stratal surfaces and constrain the 2-D and 3-D outcrop models and for 3-D visualization of the outcrops. and combination stochastic-deterministic models (Novakovic et al. Robinson and McCabe. 3-D models of the point-bar deposit were also constructed that represent similar scenarios of stratigraphic and petrophysical heterogeneity as in the 2-D models. 2002). 2003). Many of these studies investigate stratigraphic architecture (i. 2001).g. sand or shale body dimensions and distribution) and reservoir connectivity on a field scale (e. For this study.9 ft) with a thickness 5 cm (2 in. the reservoir volume swept at BTT and sweep efficiency refer to the reservoir area (or volume for 3-D models) and per- The 2-D and 3-D outcrop model frameworks were defined and built using high-resolution LIDAR data (Figure 9). Stephen et al. 1993). Patterson et al. 2001. 2001). Hartkamp-Bakker and Donselaar.and three-dimensional modeling of the main point-bar deposit in Coal Canyon (Figure 5) was conducted to quantify and evaluate intermediate-scale lithological and petrophysical heterogeneity within the fluvial point-bar deposit using deterministic and stochastic techniques. Porosity and permeability models were used to calculate rock and pore volumes and conduct single-phase streamline flow simulations. Leeder. Edwards et al. 1973.900 model cells. Because 2-D geologic modeling and flow simulation can commonly be overly pessimistic in predicting reservoir behavior and performance (Li and White. Bridge and Mackey.e. reservoir volume swept at BTT. Jackson.MackeyandBridge. detailed (centimeter-scale) measured sections (Figure 7). Tyler et al. LIDAR uses a low-energy laser and sensitive receiver to acquire detailed digital-elevation models of the outcrop (Jennette et al. 1978. In this study. OUTCROP RESERVOIR MODELING 2-D and 3-D Stratigraphic Frameworks High-resolution geologic modeling of outcrops is useful to identify the types and scales of lithologic and petrophysical heterogeneity that affect fluid-flow behavior and rock and pore-volume estimates (Joseph et al. 1993. 2003). 1994. the overall impact of the model size is less important than the degree of internal heterogeneity that is modeled. 1976.. data on modern and exhumed ancient point bars were used (Puigdefabregas.. 1978. which served as constraints for petrophysical modeling. Willis and White. The streamline simulations compare the relative effects of lithology.) were used to capture the observed lateral and vertical heterogeneity... and petrophysical heterogeneity on breakthrough time (BTT ). the outcrop framework is based on the northwest exposure of the point-bar deposit. zone). 1979. including process-based (Allen. respectively. 1998. For the general 3-D point-bar shape. Cell dimensions of 1. 1980. 1998. 1982. For the 2-D models. 2000. 1995..how the 3-D reservoir models were constructed (discussed below). Bridge and Leeder.. The model framework was built using an xy-regular grid format that maintained a constant cell width and length. 2002. centage of the total reservoir area (or volume). that are contacted by injected water as determined at the time of first water breakthrough in the producing well of the model. MacDonald et al. Light-ranging and light-intensity data are combined to generate 3-D outcrop images with near-zero spatial distortion and centimeter-scale vertical and meter-scale horizontal resolution. 1978. 2000... Dalrymple.e.. Sedimentologic data collected from outcrop were used to build deterministic lithology and grain-size models. 1993. Sullivan et al. The 3-D point-bar model framework consists of surfaces that represent the top and base of the point bar and internal lateral-accretion surfaces constrained to observations and data from all three outcrops and their respective LIDAR data. 1997). Four methods of modeling petrophysical properties were investigated (two deterministic and two stochastic).7  4.stochastic(Hirstetal.. 2003). and interpretations of the outcrops in Coal Canyon. 1983).

Figure 9. (B) northwest. 1038 Outcrop Modeling of Point-Bar Heterogeneity. Paired ground-based LIDAR (upper) and photomosaic (lower) images of the (A) southwest. and (C) east outcrops within the Coal Canyon study area. Williams Fork Formation .

The geometry and character of the lateral-accretion surfaces were interpreted from the LIDAR data and represent what is observed in outcrop. Surface topography of the point bar was based on published experimental data (Willis.7-ft) cells.2 ft).) thick. Those zones are further subdivided into thin layers that are each 5 cm (2 in.1 m (25.) thick. 980  2460 ft). An xy-regular (constant cell width and length) 3-D model grid with 810.Figure 10.6  13. All other zones have cell thicknesses of 0. Each 5-cm (2-in. Cell dimensions are 7.5 ft) with variable thickness. so that the thickness of the shale as observed in outcrop is accurately represented. Twodimensional stratigraphic framework of the point bar exposed in the northwest outcrop. Figure 11.13-m (4.5 m (20 in.8  4.).5  1. Those zones that define shale drapes on lateral-accretion surfaces have cells that are only 5 cm (2 in.8  4. Perspective view of the LIDAR outcrop elevation model (contoured surfaces) and the 3-D point-bar model (3-D grid) relative to the LIDAR data.5 m (8. All zones within the model were top truncated to represent the erosional character of the upper part of the accretionary units. The view is looking northward. 1989) that show that the maximum variation in pointbar topography and the thickest part of the deposit occur at the bend apex. and 81 layers.9  3.1 m (25. Thirteen subgrids or zones.6  13. 1039 . Cell dimensions are 7. Pranter et al.5 ft) aerially. each indicated by different gray shading. 100 columns.000 cells was built with 100 rows. subdivide the point-bar reservoir. and with lateral dimensions discussed above (300  750 m. Nine zones were generated using the defined stratigraphic surfaces.) layer is divided into 1. Contour interval is 2.

1 to 1%.e. Four methods were used to distribute petrophysical properties for the 2-D models: (1) uniform distribution (deterministic).. The algorithm used to generate the stochastic porosity models is a fast Fourier transform. however. 7). Porosity Modeling Average porosity for productive sandstone reservoirs within the Williams Fork Formation ranges from 6 to 12% (Cumella and Ostby. 1987). (2) stochastic simulation. Figure 12C). The other lithology scenarios and corresponding models were built using either continuous shale drapes on lateral-accretion surfaces that could potentially compartmentalize the reservoir into separate hydraulic flow units (lithology model 2.. and (4) a method related to grain size (deterministic). The measured sections show two scales of fining upward in grain size: the first at the scale of the overall point-bar deposit and the second within each lateral-accretion unit. In the absence of porosity data linked to grain size for the Williams Fork Formation outcrops of this study. For the porosity models generated using stochastic simulation with an imposed vertical trend. variability in grain size within the point-bar models correlates to facies transitions observed in outcrop. This scenario assumes that there is no internal lithologic variability within the point bar. The surrounding units (i. the grainsize distribution within the point-bar deposit was also modeled as an additional scenario for comparison to lithology model scenarios. (2) stochastic simulation.g. The most simplistic lithology scenario (lithology model 1. Because grain size has a significant control on petrophysical properties (Beard and Weyl. porosity or permeability) and then performs kriging on the unconditional simulated field (Ripley. the drapes are restricted to the topographically higher parts of the accretion surfaces. a basic linear transform was used to relate measured sandstone grain size from outcrop to average porosity Outcrop Modeling of Point-Bar Heterogeneity. Lithology scenario 3 is the closest approximation of the actual lithologies or bounding layers present in the point-bar deposit of this study (Figures 5B. 1987). Because of its negligible effective pore volume (noneservoir rock) compared to the sandstone. For the 3-D porosity and permeability models of the entire point-bar deposit. This porosity range for sandstone was honored in the porosity models except for the cases based on a uniform distribution in which sandstone porosity was set to 9% (the median value of productive sandstones within the Williams Fork Formation). In general. Figure 12B) or discontinuous shale drapes that could create tortuous paths for fluid movement (lithology model 3. which performs the convolution of a white noise field with a filter to create the correct correlation structure (Ripley. and three lithology models of the entire point bar were generated for 3-D modeling. For scenario 3. the other scenarios have been observed in other point-bar deposits within Coal Canyon. Petrophysical Models Deterministic and stochastic methods were used to model porosity and permeability using the lithology and grain-size models as constraints. shale was assigned porosity values that ranged from 0. crevasse splay and flood-plain mudstones) were modeled as homogeneous sandstone or shale layers. The procedure involves unconditional simulation of the spatial variable (e. Deterministic methods were used to model the distribution of sandstone and shale by simply assigning values of lithology to the layers or zones of the corresponding stratigraphic framework. These common modeling methods were used to produce different levels of detail (different petrophysical scenarios). Shale drapes on lateral-accretion surfaces represent the most significant stratigraphic features that could act as flow baffles or barriers. 1973). These trends were modeled (Figures 7. the properties were modeled using (1) uniform distribution. and sand-on-sand contacts exist near the toe of the accretion surfaces. 12D) using linear interpolation and assuming isotropic conditions. and (3) a method related to grain size. The different 2-D and 3-D lithology models thus represent the internal heterogeneity of the point-bar deposit of this study as well as other point-bar deposits observed within Coal Canyon. three different lithology models were constructed of the northwest outcrop for 2-D modeling (Figure 12A–C). (3) stochastic simulation with an imposed vertical trend. 2003). Figure 12A) consists of 100% sandstone within the point bar. a linear function in stratigraphic depth was used to impose a decreasing-upward porosity trend within the point-bar model. Williams Fork Formation .Lithology and Grain-Size Models Using the 2-D and 3-D modeling frameworks. The grainsize models most likely provide more realistic distri1040 butions of internal heterogeneity when compared to the previously described lithology models because grain size is commonly inherently linked to petrophysical properties.

footto-foot samples) to obtain petrophysical measurements was not conducted because of limited accessibility of the outcrop and for safety reasons. accordingly. 1041 . Based on information from the measured sections and field samples. The shale drape thickness is exaggerated in the figure for visibility.g. The lithology scale is applicable to (A–C). (A) Lithology model 1 (2D-LM1) is a homogeneous point-bar scenario. This lateral range was consistently used with a spherical function and zero nugget effect in all stochastic petrophysical modeling. Nonetheless. The vertical correlation length is generally a fraction of the lateral or Pranter et al.Figure 12. the lateral correlation length or range for porosity was estimated to be 20 m (66 ft). Within the point-bar sand body. (D) Grain-size model (2D-GS) based on measured section data. this type of information is essential to quantify lateral vari- ability of properties at the scale of interest and is an important input to reservoir modeling. grain size decreases upward on the scale of individual accretion units and throughout the entire point-bar deposit. Twodimensional lithology and grain-size models of the northwest outcrop.. of productive sandstone reservoirs within the Williams Fork Formation: f ¼ 0:4167 ðgrain size in millimetersÞ  0:0037 where f is porosity expressed as a fraction. This function was used to convert the grain-size models to porosity. (B) Lithology model 2 (2D-LM2) has shale drapes that extend across the entire length of each lateral-accretion surface. Detailed lateral sampling of the outcrop (e. (C) Lithology model 3 (2D-LM3) has shale drapes present only on the topographic highs of the lateral-accretion surfaces. This length corresponds to approximately one-fourth of the horizontal dimension of the dominant lateral-accretion bedding of the point-bar deposit (perpendicular to the paleoflow direction).

have the most random distribution of porosity. the models based on stochastic simulation with a linear trend have the highest average values of porosity. and one based on grain size (3D-GS-TR). Three 2-D porosity models based on stochastic simulation are designated as 2D-LM1-S.3 ft) was consistently used in all stochastic petrophysical modeling.1 – 2 md (Cumella and Ostby. coal seams. Similarly. 3D-LM2-S. These models. GS = grain size and TR = transform. and 2DLM3-S (where model 2D-LM1-S is constrained to lithology model 1 and so forth. outcrop-based permeability values for the Williams Fork Formation are also anomalous because they are affected by different burial histories compared to equivalent subsurface reservoirs and by meteoric diagenesis (German. 2D-LM2-S. and 2D-LM3. respectively. 1992). Figure 13D).. seven 3-D porosity models were generated: three based on a uniform distribution (3D-LM1U. 3D-LM3-S). and S = stochastic). and 2D-LM3-U (2D = two-dimensional. Williams Fork Formation . 3D-LM2-U. including isolated. The decreasing-upward trend in porosity follows the grid-cell layering. Ten 2-D porosity models were generated based on the four petrophysical distributions and conditioned to the three lithology models and the grain-size model (Figure 13). and ST = stochastic simulation with a trend). 2D-LM2-U. the following transforms were used to convert porosity (f) to permeability (k) for the 2-D and 3-D models: f k ¼ 10:811 e0:3199 ðfor sandstoneÞ f k ¼ 0:05 e0:3199 ðfor shaleÞ From the single porosity model based on grain size. Permeability data from a Middle Jurassic fluvial reservoir of the Brent Group (Ness Formation) in the northern North Sea were used. Three 2-D porosity models with a uniform distribution are designated 2D-LM1-U. 3D-LM3-U). 2003). These models represent the simplest spatial distribution of porosity of all model scenarios that were evaluated. in comparison to the other three scenarios. fining-upward fluvial channels. Streamline Simulations To assess the relative effect of lithologic and petrophysical heterogeneity on fluid flow within the pointbar reservoir. Based on these data (Figure 14). and the lowest average porosity is associated with the grain-size model. 2006). 2D-LM2-ST. the Ness Formation was deposited as a complex of coastal sediments. to make the outcrop modeling results useful and applicable to more conventional fluvial reservoirs and to efficiently run singlephase streamline flow simulations using the lithology and petrophysical modeling results. The 2-D grainsize model was transformed to create a porosity model based on grain size as discussed previously (2D-GS-TR.horizontal correlation length in most sedimentary formations. In general. Three 2-D porosity models based on stochastic simulation with a decreasing-upward porosity trend are designated as 2DLM1-ST. permeability data from a conventional (non – tight-gas sandstone) fluvial reservoir analog were used to model permeability. Datta-Gupta Outcrop Modeling of Point-Bar Heterogeneity. U = uniform distribution) and are constrained to lithology models 1 to 3 (2D-LM1. causing the stratigraphically lower parts of the point bar to have the highest porosities. Similar to the Williams Fork Formation. which is on the order of 1042 0. The permeability values of Williams Fork Formation reservoirs are therefore unique because they are associated with tightgas sandstone reservoirs. Explanations of the streamline simulation methods that were used are provided by Pollock (1988). In addition. three different permeability models were produced with values that correspond to the distributions of shale drapes as in the three lithology models (i. a vertical correlation length of 1 m (3. without shale drapes and continuous and discontinuous shale drapes). Core plugs from productive sandstones and shales of the Brent Group were analyzed for porosity and permeability by Stiles and Hutfilz (1992) and later compiled by Nelson and Kibler (2003). and 2D-LM3-ST (where model 2D-LM1-ST is constrained to lithology model 1 and so forth. However. the Williams Fork Formation produces mainly from fracturerelated permeability in the subsurface caused by the very low matrix permeability. three based on stochastic simulation (3D-LM1-S. Based on measured-section data.e. Permeability Modeling This study addresses intermediate-scale fluvial reservoir heterogeneity based on the Williams Fork Formation because its stratigraphic architecture and internal lithologic heterogeneity is characteristic of fluvial reservoirs in various basins and settings. These permeability data are used because they are characteristic and representative of many conventional fluvial reservoirs. 2DLM2. single-phase streamline simulations were conducted using the 2-D and 3-D outcrop models. Figure 12). and lagoonal mudstones (Richards. LM = lithology model number. Because the focus of this study is not the petrophysical properties of tight-gas reservoirs.

and sweep efficiency reveal that lithologic heterogeneity has a greater effect on recovery efficiency than petrophysical heterogeneities (disregarding the 2-D grain-size–based model. The results of the 2-D flow-simulation runs and analysis of variance statistical tests that compare BTT. (D) Model 2D-GS-TR is based on a transformation (TR) of the grain-size model (2D-GS of Figure 12D) to porosity. (A) Model 2DLM1-U is based on a uniform distribution (U) and constrained to lithology model 2 (2D-LM1 of Figure 12A). a streamline approximates the trajectory of a moving fluid particle through the reservoir (Pollock. (B) Model 2D-LM2-S is based on stochastic simulation (S) and constrained to lithology model 2 (2D-LM2 of Figure 12B). 16). pore volume swept at BTT. As an example. and King (1995). which is discussed below). Examples of 2-D outcrop porosity models. In general. for the models based on stochastic simulation. the continuous (2D-LM2-S) Pranter et al. Twelve 2-D streamline simulations were run using the porosity and permeability models conditioned by the different lithology models. 1043 . The simulations were compared in regard to BTT. Both wells penetrated the entire thickness of the model. (C) Model 2D-LM3-ST is based on stochastic simulation with an imposed trend (ST) and constrained to lithology model 3 (2D-LM3 of Figure 12C). All 2-D streamline simulations involved an injection well on the south (left) side of the model and producing well on the north (right) side. pore volume swept at BTT. and sweep efficiency (Figures 15.Figure 13. only the main point-bar sandstone was open to flow in the injection and production wells. 1988). and Ponting (1998). however.

3% decrease in the pore volume swept at BTT and sweep efficiency. The various 2-D models based on grain size were all produced in a similar way. and sweep efficiency of the different grain-size models are also similar (Figure 15A–C). no preferential paths of fluid flow are established (Figure 16A. compared to the models without shale drapes (Figure 15B. 1996). absolute BTT for well pattern 1 was significantly Outcrop Modeling of Point-Bar Heterogeneity. Well pattern 2 is comparable to the well positions used in the 2-D outcrop modeling. is observed (Figure 15B. 16C). This lithologic heterogeneity associated with the discontinuous shales (Figure 12C) is representative of the heterogeneity that is present in many point-bar deposits because shale 1044 drapes are commonly only preserved on topographic high areas of the lateral-accretion surfaces (Miall.9% decrease is observed in the pore volume swept at BTT and sweep efficiency. pore volume swept at BTT. Data are based on core plugs from productive sandstones of the Brent Group (Ness Formation) (Stiles and Hutfilz.. compared to the models without shale drapes (Figure 15A).5 and 2. respectively. the sweep efficiency values are greater than 80% (Figure 15A. Porosity-permeability crossplot and the best-fit porosity-permeability transform used in 2-D and 3-D petrophysical modeling. respectively. The 3-D streamline simulations were also evaluated in terms of BTT. Although the BTT values are different between models based on lithology models 1 and 2.7 and 30.3% increase and 24. C). Williams Fork Formation . Well pattern 2 is oriented so that the injected fluid flows from south to north. parallel to the paleoflow direction (Figure 17A–C). For the 3-D streamline simulations. C). Because these baffles are continuous. C). respectively. perpendicular and parallel to the paleoflow direction). C. Well pattern 1 is oriented so that the injected fluid flows from east to west. the resulting BTT. and sweep efficiency (Figure 15D–I). However. Nelson and Kibler. 2003).Figure 14. The 2-D streamline simulations of the models based on grain size exhibit 67–267% shorter BTT (Figure 15A) and 44–57% lower sweep efficiency (Figure 15C) than the other models.g. and as expected. only a 5. The discontinuous shales act mainly as permeability baffles where they are present and force the flow downward toward the base of the point bar where shale drapes are not present. perpendicular to paleoflow (Figure 17D– F). Therefore. for the continuous-shale-drape model (2DLM2-S). Given the greater distance between wells for well pattern 1. a 31. 1992. The continuous shales act as baffles to flow. thus reducing both the BTT and sweep efficiency as compared to the other cases (Figures 15A. pore volume swept at BTT. two well patterns were used to compare the relative effect of fluid flow with respect to the orientation of the architectural framework within the point bar (e. the petrophysical variability among just the grainsize models is similar. B). and discontinuous (2D-LM3-S) shale drapes on lateralaccretion surfaces within the point bar cause a 79. For the discontinuous-shale-drape model (2D-LM3-S). causing an increase in BTT compared to the homogeneous case.1% decrease in BTT.

LM3 = lithology model 3 (light-gray bars). GS = distribution based on grain size. uniformly inhibit fluid flow. for a given lithology constraint. the magnitude of the differences in the measured parameters among the different scenarios (excluding the models based on grain size) is negligible (Figure 15D – F). In general. the discontinuous shale drapes of lithology model 3 act as baffles. BTT = breakthrough time. for the 3-D models. Histograms of streamline simulation results for 2-D (A – C) and 3-D (D – F ) models. and sweep efficiency was lowest for the models based on the grain-size distribution (Figure 15D – I). Model results are coded as follows: LM1 = lithology model 1 (black bars). In contrast to the 2-D models. and result in relatively longer BTT compared to 3D-LM1 or 3DLM3. shale drapes act as lateral permeability baffles. I). This is especially true for the models with continuous shale drapes (3-D lithology model 2. 3D-LM2). and lead to similar BTT and sweep efficiency compared to the uniform Pranter et al. ST = stochastic simulation with a vertical trend. In the case of 3D-LM2. LM2 = lithology model 2 (gray bars). longer for all cases than for well pattern 2 (Figure 15D. For well pattern 1 (parallel to paleoflow). The fining-upward trends in grain size produce the most distinct and focused fluid-flow pathways and result in shorter BTT and lower sweep efficiency (Figure 15D–F). S = stochastic simulation.Figure 15. as in the 2-D models. focus fluid flow downward to the base of the point bar. U = uniform distribution. the lithologic distribution has a more significant control on the resulting streamline parameters compared to the petrophysical distribution. For well pattern 2 (perpendicular to paleoflow). the continuous shale drapes of lithology model 2 actually cause the fluid front to spread out laterally and contact more of the reservoir (Figure 17E). The shale barriers present on the topographic highs of the accretion surfaces in lithology model 3 (3D-LM3) do not focus or confine the fluid flow in the direction parallel to paleoflow. BTT was shortest. Simulations based on lithology model 2 have 42 – 53% longer BTT and 41 – 52% higher sweep efficiency compared to the other models (Figure 15G. G). However. 1045 .

For comparison. The spatial distribution of petrophysical properties. thus. form isolated reservoir compartments. Time from injector (TFI) is displayed at BTT for each model. fluid-flow behavior. The results of this outcrop analog study show how lithologic heterogeneities (e. point-bar deposits (and other related sand bodies) are commonly disconnected or in limited fluid-flow communication 1046 and.. IMPLICATIONS FOR RESERVOIR MODELING AND DEVELOPMENT For fluvial reservoirs. the input models for the streamline simulations correspond to the same models as shown in Figures 12 and 13. and reservoir Outcrop Modeling of Point-Bar Heterogeneity. 17D –F). laterally discontinuous shale drapes on accretion surfaces. and affect recovery efficiency under different development scenarios.g. grain-size variability) within a point-bar deposit of a fluvial reservoir can act as baffles or barriers to fluid flow to compartmentalize a reservoir. U = uniform distribution. control fluid movement. case. GS-TR = transform based on grain size. especially those associated with low net-to-gross meandering streams. The TFI is the amount of time it takes a particle of water to move from the injector well to that location in the model given the boundary conditions. Within these types of reservoir elements. S = stochastic simulation. Williams Fork Formation . LM1 – 3 = lithology models 1–3. internal lithological and petrophysical heterogeneities commonly compartmentalize the reservoir. BTT = breakthrough time. Streamline simulation results for select 2-D model examples. but relatively shorter BTT and lower sweep efficiency compared to models with continuous shale drapes (Figures 15G.Figure 16. ST = stochastic simulation with a vertical trend. I.

models in (B) and (E) were conditioned using lithology model 2. Maps of streamline simulation results for select 3-D model examples.Figure 17. Pranter et al. 1047 . Models in (A – C) were run using well pattern 1. where fluid flow is oriented parallel to the paleoflow direction within the point bar. where fluid flow is oriented perpendicular to the paleoflow direction within the point bar. Models in (D – F) were run using well pattern 2. Models in (A) and (D) were conditioned using lithology model 1. Time from injector (TFI) is displayed at BTT for the petrophysical models based on stochastic simulation. The TFI is the amount of time it takes a particle of water to move from the injector well to a location in the model given the boundary conditions. and models in (C) and (F) were conditioned using lithology model 3.

g. HartkampBakker and Donselaar. 2000). each point-bar deposit acts as a separate reservoir element. 1978. horizontal wells drilled perpendicular to the paleoflow direction would allow multiple reservoir compartments to be contacted (in the case of continuous shale drapes) and more uniform sweep and higher sweep efficiency to be achieved (in the case of discontinuous shale drapes) relative to a vertical well. which exhibit shorter BTT and lower sweep efficiency compared to the other models because of the fining upward in grain size. 1986. The point-bar deposit of this study exhibits a vertical succession of facies from medium. Within each point-bar sand body (reservoir compartment). Distinct lateral-accretion surfaces within the point-bar deposit commonly have thin shale drapes present on the updip part of these surfaces. 1991. Two. a relatively high well density (e. trough cross-bedded sandstone to fine-grained ripple-laminated sandstone. and with discontinuous shale drapes is negligible. lithologic heterogeneity and directional permeability (petrophysical anisotropy) of varying magnitudes associated with cross-stratified sandstone. but only moderate differences in the pore volume swept at BTT and sweep efficiency.. Therefore. However. rippled and laminated sandstone. The shale drapes do not focus or confine the fluid flow in the direction parallel to paleoflow. average-property values (deterministic methods).. Tyler and Finley. Grain size decreases vertically at the scale of individual accretionary units and for the entire point bar. for the models based on grain Outcrop Modeling of Point-Bar Heterogeneity.g. Because finingupward grain-size trends can partially compartmentalize the reservoir vertically and reduce sweep efficiency. therefore. Jones et al. (1987) and Anderson (2005) also suggest that maximum permeability in trough cross-bedded sandstone is parallel to the trough axis and.g.performance differ between models based on lithology. for a given well spacing. the reproduction of internal lithologic variability.and coastal-plain settings. horizontal or deviated wells might be preferred depending on the magnitude of grain-size variability. isolated point-bar sand bodies) encased in floodplain mudstones and coals. 1993. therefore. The lower Williams Fork Formation is a relatively low net-to-gross ratio interval. For the 3-D models with producer-injector well pairs aligned parallel to the paleoflow direction. 1976. and it has an associated fining-upward grain-size profile. 660– 1320-ft] spacing or less) is necessary to intersect these individual reservoirs. Given the geometry and dimensions of the pointbar sand bodies of the lower Williams Fork Formation and similar low net-to-gross ratio systems. This is also the case for the models based on grain size. producer-injector well pairs should be aligned parallel to the paleoflow direction to maximize sweep efficiency. the magnitude of the differences in BTT and sweep efficiency among the models without shale drapes. These deposits include relatively isolated but internally heterolithic sandstones (e. excluding the models based on grain size. Continuous and discontinuous shale drapes on the lateral-accretion surfaces within the point bar cause significant differences in BTT. 1992. parallel to the paleocurrent direction. 1048 CONCLUSIONS The Williams Fork Formation of the western Piceance Basin consists of interbedded sandstone. shorter BTT parallel to paleoflow is expected...and three-dimensional modeling results indicate that the effect of internal lithologic heterogeneity on fluid flow varies.. and coal that were deposited in alluvial. Willis and White. shale. Williams Fork Formation . Richardson et al.. Thomas et al. and associated petrophysical properties within modeled point-bar deposits (reservoir-model objects) are important depending on the character and degree of internal heterogeneity that exist (e. Jordan and Pryor. The results suggest that when producing a fluvial reservoir with discontinuous shale drapes. it would be favorable to align production-injection well pairs such that the displaced fluid flows parallel to the paleoflow direction. and distributions based on stochastic simulation. Anderson 2005). depending on the type and extent of the heterogeneity and the point-bar orientation relative to development wells. with continuous shale drapes. Similar to the 2-D models.to upper finegrained. if lithological variability exists and vertical wells are used. The discontinuous shales act mainly as permeability baffles where they are present and force the flow downward toward the base of the point bar. This is especially true if the shale drapes are essentially impermeable. 1987. 10–40-ac [201–402-m. Lasseter et al. 1987. Reservoir connectivity between the point-bar sand bodies is generally low. and other facies influence fluid movement (Jones et al. If shale drapes on lateral accretion surfaces exist (either continuous or discontinuous).. Therefore. depositional trends. The continuous shales inhibit flow along their entire length and cause an increase in BTT compared to the homogeneous case. Jackson. HartkampBakker and Donselaar. thus reducing both the BTT and sweep efficiency as compared to the other cases. 1993. observed grain-size trends.

C. v. Peterson. 21.. 18.. Kier. with comparison to the Piceance Basin subsurface: The Mountain Geologist. Carlston. p. Mackey. Analysis of fluvial sandstone-body characteristics and architecture in a high net-to-gross system. and G. and S. Donnell. A theoretical study of fluvial sandstone body dimensions... S. 71. v. J. D. R. T. p.. Colorado. S. S. REFERENCES CITED Allen. J. Datta-Gupta. Colorado: Implications for reservoir description: The Mountain Geologist. Ethridge. 171 – 193. v. Colorado: U. p.. 115 – 122. Architecture of crevasse splay and point-bar bodies of the nonmarine Iles Formation north of Rangely. and assessment.. L. in A. Cole. S. T. For a well pattern oriented perpendicular to paleoflow. 577 – 586. 85 – 107. the continuous shale drapes also cause the fluid front to spread out laterally and contact more of the reservoir. A.. 2001. it would be favorable to align production-injection well pairs such that the displaced fluid flows parallel to the paleoflow direction. and lead to relatively shorter BTT and lower sweep efficiency compared to the other models. A. eds. 26. v. M. L. The relation of free meander geometry to stream discharge and its geomorphic implications: American Journal of Science.. Allu- vial sedimentation: International Association of Sedimentologists Special Publication 17. Combining sequence stratigraphy. Cumella. S. Fluvial sedimentology: Canadian Society of Petroleum Geologists Memoir 5. Utah. Stratigraphic architecture and reservoir characteristics of the Mesaverde Group. ed. Cumella. 42. 2004. 1961. Cole. B. Q. F. Allen. K. Unlike the 2-D model cases. 213 – 236. A. p.. and Garfield and Mesa counties. D. M. Puigdefabregas. R. 1965. v.. J. and S. Colorado: Quarterly of the Colorado School of Mines. and R. As a result. Reeside. 2002. G. Colorado: Master’s thesis. the discontinuous shale drapes act as baffles. the pore volume swept at BTT and sweep efficiency are greater. D. 2006. no. France): AAPG Bulletin. 1973. 1978.. and southern Danforth Hills coal fields. Paleochannel geometry and flow patterns determined from exhumed point bars in north-central Texas: Journal of Sedimentary Petrology. 35. Smart. Schumm. Leeder. Geological Survey Bulletin. and J. 385 – 442. Laramie. S. eastern Piceance Basin. v. King. in S. W. Emery. v. M. ed. R.. 1977. S. 4. P. German. Collins. eds. R. J. 1983. 263. in M.S. September 6 – 8. Geological Survey Professional Paper 332. Coal deposits of the Carbondale. Fisher.. B. 2. no. Sand body architecture in the lower Williams Fork Formation (Upper Cretaceous). A simulation model of alluvial stratigraphy: Sedimentology. Beard. 7. B. Colorado. v. R. Gustason. J. limitations. v. M.. J. 109 – 122.. 1082-L. Parpant. Boulder. v. Wyoming. A. in A. P. northwestern Colorado: U. p. 1979. J. upper Pranter et al. and S. Bridge. D. Colorado. 349 – 369. North Wales: Proceedings of the Yorkshire Geological Society. Geology of the basincentered gas accumulation. 1. 1995. Bacciana. Simulations based on the continuous shale drapes have the longest BTT and highest sweep efficiencies because they act as lateral permeability baffles and uniformly inhibit fluid flow.. A semi-analytical approach to tracer flow modeling in heterogeneous permeable media: Advances in Water Resources. focus fluid flow downward to the base of the point bar.. Boyles. Piceance Basin. Cretaceous and Tertiary formations of the Book Cliffs. Flint and I.size. Tertiary geology and oil-shale resources of the Piceance Creek Basin between the Colorado and White Rivers. p.. 617 – 644. If continuous or discontinuous shale drapes exist. The geological modelling of hydrocarbon reservoirs and outcrop analogs: International Association of Sedimentologists Special Publication 15. eds. and S. 80 p. J. D. R. Fluvial reservoir architecture in the Statfjord Formation (northern North Sea) augmented by outcrop analogue statistics: Petroleum Geoscience. and S. 1049 . 319 – 336. 703 – 722. Olsen. C.. Coal Canyon. D. 2002. I. in K. 1976. G. M. Colorado. Anderson. D. Studies in fluviatile sedimentation: An exploratory quantitative model for the architecture of avulsioncontrolled alluvial suites: Sedimentary Geology. The sedimentation and paleogeography of the Old Red Sandstone of Anglesey. v. Piceance Basin 2003 guidebook: Rocky Mountain Association of Geologists.. B. p. Bridge. Miall.. no. Collinson. Vertical sequence and sand body shape in alluvial sequences. p. S. Numerical modeling of heterogeneity within a fluvial point-bar deposit using outcrop and LIDAR data: Williams Fork Formation.. northwest Colorado: Field trip 2 in conjunction with the 2002 Rocky Mountain Sectional Meeting of the AAPG. a single horizontal well drilled perpendicular to the paleoflow direction would allow multiple reservoir compartments to be contacted and more uniform sweep and higher sweep efficiency to be achieved relative to a vertical well. southern Piceance Basin. and P. Olsen. p. Peterson. v. 42. D. C. Grand Hogback. geostatistical simulations. S. Dalrymple.. 1965. Stratigraphic architecture and reservoir characterization of the Mesaverde Group... Mackey. Lemouzy. 2005. Similar to the 2-D models. 234 p. Erikkson. Anderson. the lithologic distribution has a more significant effect on BTT and sweep efficiency as compared to the petrophysical distribution. p. Bryant. 1993a. 80 p. if vertical wells are used. 129 – 147.S. 53. 1998. 57. These results suggest that when developing a fluvial point-bar reservoir with internal discontinuous shale drapes. eds. Miall. Anderson. no. 545 – 568. Ellison. Jr. and M. University of Colorado. p.. A. p. R. 2003. p. 2003. 1261 – 1270. v. and B. and Grand counties. Carbon. and production data for modeling a fluvial reservoir in the Chaunoy field ( Triassic. Marzo and C. Desaubliaux. Eschard. p. A. and M. J. Cumella. S.. 9 – 25. Piceance Basin. 138 p. M. Edwards. p. 2005. Cole. D. Weyl. 1960. 82. 1978. Cumella. the fining-upward trends produce the most distinct and focused fluid-flow pathways and result in shorter BTT and lower sweep efficiency. P.. in K. Erdmann. Influence of texture on porosity and permeability of unconsolidated sand: AAPG Bulletin. p. Piceance Basin 2003 guidebook: Rocky Mountain Association of Geologists. C. and D. Ostby. 139 – 185. 4. M. E. Reconstructing paleochannel morphologic and flow characteristics: Methodology. R. p. 835 – 891. 864 – 885. Fluvial sedimentology: Canadian Society of Petroleum Geologists Memoir 5. Bridge. 1993b. A revised alluvial stratigraphy model. J.. p. and D. and D. K. D.

S. A. 1960. D. Kerans.. p. p. B. Kirschbaum. 1985. central Spain. Doyle. 27.gov/dds/dds-069/dds-069-b /REPORTS/Chapter_12. v. A study of the Cretaceous – Tertiary unconformity in the Piceance Creek Basin. G. Hoak. v. Joseph. 1998. 1986. Jackson. Anderson.. p.. Halland. E. A. Carroll. 25 p. Mackey. Fractured reservoirs: Characterization and modeling: Rocky Mountain Association of Geologists guidebook. Johnson. Colorado: Master’s thesis. M. 1993. 6. p. L. University of Colorado. Waggoner. comp. K. Stochastic modeling of incised valley geometries: AAPG Bulletin. SPE Paper 13052. no. Deep-Water Reservoirs of the World: Gulf Coast Section SEPM Foundation 20th Annual Research Conference.gov/dds/dds-069/dds-069-b/REPORTS /Chapter_7.. Geological Survey Investigations Series I-2764.S.. J. p. The geological modelling of hydrocarbon reservoirs and outcrop analogs: International Association of Sedimentologists Special Publication 15.S. Blomquist. J.. A. P. R. M. and C. 175 p. D. Utah and Colorado: U. Nelson. 2003. D. W. 3 – 29. Tyson. Preliminary evaluation of lithofacies models for meandering alluvial streams. R. Bryant. Donselaar.. 1997. and D. http://pubs. in L. ed. J. 21 p. v. http://pubs. S. 1601 – 1624. 5. 97– 107. R.. Geological Survey Digital Data Series DDS-69-B. Kirschbaum. S. Determination of widths of meander-belt sandstone reservoirs from vertical downhole data. L. Lasseter. Utah and Colorado: U. 1995. History of the Piceance Basin from latest Cretaceous through early Eocene and the characterization of lower Tertiary sandstone reservoirs. 5 – 18. Colorado: The underlying Ohio Creek Formation (Upper Cretaceous) redefined as a member of the Hunter Canyon or Mesaverde Formation: U. Hartkamp-Bakker. 2002... p. 1016 – 1040. Bryant. 6 p. C. T. Geostatistical models for shales in distributary channel point bars (Ferron Sandstone. Fluvial sedimentology: Canadian Society of Petroleum Geologists Memoir 5. and D. E. 1978. eds. p. 76. p. in P. J. 2003. Fluvial sedimentology: Canadian Society of Petroleum Geologists Memoir 5. Piceance Creek Basin. C. Flint and I. Coleman. Statfjord field. C. MacDonald. The Annot Sandstone outcrops (French Alps): Architecture description as input for quantification and 3-D reservoir modeling. and F. R. Lake and H. 1787-E. Leopold. A. B. Outcrop Modeling of Point-Bar Heterogeneity. W. and R. in T. T. D. C. B. MacDonald. W. Johnson. 71. p. Reservoir heterogeneities and their influence on ultimate recovery. 1482-B. Jones.. Geological Survey Open-File Report 03-420. A. 3. Jordan.usgs. Nelson. p. 543 – 576. Depositional model of point bars in the Lower Wabash River: Journal of Sedimentary Petrology. in USGS Uinta-Piceance Assessment Team. p. J. H. Miall. 1993. and M. with special reference to channel deposit density and interconnectedness. Nanson.. 682 – 697.. D. p. R. 67 – 102. Kibler. deHaan. C.. 545 – 559. v. Utah and Colorado: U. A breakthrough on the outcrop: New tools and techniques to bring outcrop studies into a 3-D workstation environment: Texas Bureau of Economic Geology midyear report 2003. C. Colorado. Flores. R. 1996. 1976. Lake. R. A. W. Johnson. 157 – 168. 265 – 276. The geological modelling of hydrocarbon reservoirs and outcrop analogs: International Association of Sedimentologists Special Publication 15. Sedimentology and shale modeling of a sandstone-rich fluvial reservoir: Upper Statfjord Formation. and J. A. v. Olson. Peterson. D. and S. Jacobsen. chapter 7. Utah): From ground-penetrating radar data to three-dimensional flow modeling: AAPG Bulletin. 1977.. Colorado: AAPG Bulletin. in S. 3. R. R. J. Lake. Miall. Jennette. Point bar and floodplain formation of the meandering Beatton River. Searls. Heinz. P. p. eds. Springer-Verlag. R. M. Clark. 1995.pdf (accessed May 2007). J. 1980. Pryor. Hettinger. and W. Canada: Sedimentology. J. Geological Survey Bulletin. R. 21 – 61. Geological Survey 1050 Digital Data Series DDS-69-B. B65. and L. Academic Press. p. 27 p. Bridge. and M. Klawitter.. White.. E. and A. and S. Which subseismic heterogeneities influence waterflood performance? A case study of a low net-to-gross fluvial reservoir. T. in USGS Uinta-Piceance Assessment Team.. Kjonsvik. A. H. E. 587 – 596. Boulder. A. v... v.. and J. et al. and E. Scott. Geological Survey Bulletin. 422 – 449. Bouma. 46. D. Rosen. 72. comp. Hettinger. D. CD-ROM. 1973. 1 – 8. 2003.. 1988.Williams Fork Formation. D. Blackstock. v. v. A. p. v. N. in A. Slatt. A. Leeder. C. M. in H. 1980. northwestern Colorado: Society of Petroleum Engineers Formation Evaluation. Williams Fork Formation . no. G.S. Utah and Colorado. plus attachments. 51 p. in K. Hoak. C. Andrews. Hektoen. Jackson.. R. and L. M. no. northeastern British Columbia. H. Piceance Basin.S.pdf (accessed May 2007). 68 p. J. v. Li. A catalog of porosity and permeability from core plugs in siliciclastic rocks: U. A. Miall. 2000.. eds. The Mesaverde total petroleum system. 87. Reservoir characterization: Orlando. L. The geology of fluvial deposits: Heidelberg. 1993. J. C. J. v. 710 – 721. C. p. S. CD-ROM.. ed. and P. Falt... L. and M. eds. H.. 237 – 252. p. 82. Hierarchical levels of heterogeneity in a Mississippi River meander belt and application to reservoir systems: AAPG Bulletin.. Permeability patterns in point bar deposits: Tertiary Loranca Basin. A. R. and J. Mesaverde Group. 6. p. Fluviatile fining-upward cycles and the magnitude of paleochannels: Geological Magazine. no.. 579 – 594. K. and C. Jones. M. 1987. Stochastic modeling of fluvial sandstone bodies. and A.. western Colorado. 769 – 794. p. Jr. A. Utah and Colorado. 2003. G.. G. Plateau Creek Canyon. in S. Lawrence. Wolman. Klawitter. D. no. 1989. C. in A. p. P. Bellian. S. 7 – 31. T. L. northern North Sea: AAPG Bulletin. Stratigraphy of the Upper Cretaceous Mancos Shale (upper part) and Mesaverde Group in the southern part of the Uinta and Piceance Basins. 69. M. 6. C. T..usgs. Piceance Basin.. The geologic aspects of reservoir characterization for numerical simulation: Mesaverde meander-belt sandstone.. v. D. no. Three-dimensional model of alluvial stratigraphy: Theory and application: Journal of Sedimentary Research. Lorenz.. May. Roberts. Stratigraphy of the Upper Cretaceous Mancos Shale (upper part) and Mesaverde Group in the southern part of the Uinta and Piceance basins. P. 582 p. Leeder. Flint and I. eds. p. D. 2003. Johnson. M. R. M. A quantitative stratigraphic model for alluvium. Styzen. 77. and M. ed. Weimer. 1. Miall. Petroleum systems and geologic assessment of oil and gas in the UintaPiceance Province. SEPM CD Special Publication 28. p. and D. New developments in improved oil recovery: Geological Society (London) Special Publication 84. 1851 – 1868. 1156 – 1172.. Piceance Basin 2003 guidebook: Rocky Mountain Association of Geologists. 1992. Geologic history and hydrocarbon potential of Late Cretaceous – age low-permeability reservoirs.S... 2003...S. River meanders: Geological Society of America Bulletin. Uinta-Piceance Province. Reservoir heterogeneities in fluvial sandstones: Lessons from outcrop studies: AAPG Bulletin. no.. J. A. Prediction of fractured reservoir production trends and compartmentalization using an integrated analysis of basement structures in the Piceance Basin. Petroleum systems and geologic assessment of oil and gas in the Uinta-Piceance Province. chapter 12. Hirst. 12.. D. R. western Colorado: U. C.

D. 10 p. C. Heterogeneity of fluvial systems. J. p. 273 – 280. Central Utah coals: Utah Geological and Mineralogical Survey Bulletin. B. eds. p. G. and J. A. 44. p. T. v. H. Fluvial paleochannels. Stochastic simulation: New York. Morton. D.. 45. K. p. N. Architectural analysis of deep-water outcrops: Implications for exploration and development of the Diana Sub-basin. p. 237 p. 2002. interpretation. 2001.. Utah: Mathematical Geology. p. Goulding. R. Stratigraphy of coal-bearing rocks of Book Cliffs. M. p. A. M. 1978. 81. Proceedings of the Geostatistics Sessions of the 31st International Geological Congress. 70. 1955. Sandstone-body and shalebody dimensions in a braided fluvial system: Salt Wash Sandstone Member (Morrison Formation). Willis. 1987. Brazil: Dordrecht. J. J. P. C. R. 1531 – 1537. Bouma. The three-dimensional facies architecture of terrigenous clastic sediments and its implications for hydrocarbon discovery and recovery: SEPM Concepts in Sedimentology and Paleontology. 26. D. N.. 6. Chambers. Willis.. v. 2002. Stephen. E. L.. D. S. McCabe. and T. H. L. H. Mingjuan. v. 1963. North Sea: Journal of Petroleum Technology. v. P. M. 1996. v. outcrop observations. Corbeanu. Nance. J. Robinson. Tyler.Novakovic. Galloway. 15 – 26. Bhattacharya. D. northwest Colorado: Gas Research Institute Report GRI95/0532. Hammon. E. and case studies: AAPG Computer Applications in Geology 3. G. D. Tyler. Geologic and hydrologic controls critical to coalbed methane producibility and resource assessment: Williams Fork Formation. 123 – 179. R. western Gulf of Mexico. v. 1973.. A. Palaeochannel reconstructions from point bar deposits: A three-dimensional perspective: Sedimentology. Richardson.. Coleman. U. Mavor.. Stiles. Donofrio. v. v. Galli. D. G. The use of routine and special core analysis in characterizing Brent Group reservoirs. Rigby. Sullivan. Tyler. Kuala Lumpur. in J. 665 – 672. 857 – 893. eds. in M. 86. 66. K. J. in J. D. D. Wood. p. CalverleyRange. 1992. Quantitative outcrop data for flow simulation: Journal of Sedimentary Research. Hybrid streamline methods. Rossen. McMechan. A. D. v. M. Sharp. and S. H. Geologic modeling of external and internal reservoir architecture of fluvial depositional systems. discontinuous shales on oil recovery: Journal of Petroleum Technology. and C. John Wiley & Sons. 397 p. p. Rio de Janeiro. D. M. C. and significance: Sedimentary Geology. B. W. no. McMurray.K. 1010 – 1031. Harris. eds. 133 – 144. no. A. Ripley. v. S. The importance of fluvial processes and related reservoir deposits: Journal of Petroleum Technology. 7. p. 1991. 1972.. C. J.. Clark. K. Jensen. 41 – 52. M. D. 743 – 750. Stern. Ponting. 1988. v.... p. Weimer. 77 – 89. P.S. Tremain. description. 177 – 202. Stochastic modeling and geostatistics: Principles. Inclined heterolithic stratification — Terminology. and M. in P. p. Influence of common sedimentary structures on fluid flow in reservoir models: Journal of Petroleum Technology. Brown. J. p. 1993. Geostatistics Rio 2000. Piceance Basin. eds. J. and J. 1994. Modeling heterogeneities in fluvial domains: A review of the influence on production profiles. An introduction to the Brent Group: A literature review. S. 704 – 713. F. J. 34.. 20. B. R. 1987. The effect of small. Architectural controls on the recovery of hydrocarbons from sandstone reservoirs. Pollock. p. in Proceedings of the 1998 Society of Petroleum Engineers Asia Pacific Conference on Integrated Modelling for Asset Management. 80. and three-dimensional flow modeling for the Ferron Sandstone. Scott. Armstrong. Outcropbased stochastic modeling of turbidite amalgamation and its effects on hydrocarbon recovery: Petroleum Geoscience. 163 – 172. A. S. C. and E. Reservoir simulations developed from an outcrop of incised valley fill strata: AAPG Bulletin. Malaysia. and R. K. W. R. C. v. R. and P. Jennette. Hutfilz. v. R. 368 – 377. Lawrence. 1267 – 1291. M. Dalrymple. Kluwer Academic Publishers.. Swanson. Young. v. D... Young. v. 34. H. 53. W. 2000. J. no.. in A. SPE Paper 39756. Utah – Colorado: Geological Society of America Bulletin. A. Styzen. 1051 .. Ottmann. R. 98 – 107. eds. Pranter et al. in W. and A. no.. 5 – 17. Hydraulic effects of shales in fluvial-deltaic deposits: Ground-penetrating radar. K. G. 788 – 802. v. Geological Survey Circular 477. and D. Sedimentary facies and intertonguing in the Upper Cretaceous of the Book Cliffs. Haszeldine. Puigdefabregas. C. Bettini.. Garfield County. Schumm. geostatistics.. 36. K. J. v. SEPM CD Special Publication 28. Nelson. D. Geology of the Brent Group: Geological Society (London) Special Publication 61. Hamblin and J. 1. and G. K. G. M. Rigby and W. Thomas. p. Gardiner. R. A. p. C. in A. and A. K. Jones. 797 – 822. J. Slatt.. 1998. 1997. Donovan. 3. Semi-analytical computation of path lines for finite-difference models: Ground Water. p. Champigny. Kaiser.. M. 7. R.. 2000. D. Tyler. S. D. Recognition of ancient sedimentary environments: SEPM Special Publication 16.. N. Utah: AAPG Bulletin. Foreman. E. Giles. R. Deep-Water Reservoirs of the World: Gulf Coast Section SEPM Foundation 20th Annual Research Conference. Van Hee. J. p.. p. p.. D. methods. 5. The Netherlands. and M. J. R. Koster. p. Patterson. 1992. 2003. J. G. Svanes. H. J. and G. J. S. Hamblin. Miall and N.. p. p.. Miocene point-bar deposits in the Ebro Basin. northern Spain: Sedimentology. R. G. C. M. Weber. 9. 20. Henriquez. A. and W.. White. Yarus and R. Schumm. 2003. Stephen. G. Utah – Colorado. White. 1 – 5. 1966. eds. control on density-driven flow and transport: Environmental and Engineering Geosciences. no. Richards. Rosen. Finley. 4. D. A tentative classification of alluvial river channels: U. Smith. 757 – 766. T. 1989. 1982.. Visser. 7 – 21.